INSIDE MARANTZ AV RECEIVER

Picked up a huge Marantz receiver (SR8200) at the local donation center for not very much money.

FAULTY VOLUME KNOB

Even though the donation center has a 7-day return policy for non-working electronics, the receiver was worth more than I paid (to me) in parts alone. I quickly discovered that the volume knob was “stuck”. It is a rotary encoder type (not a potentiometer type). The volume setting would barely move when turning the knob,

Thanks to my familiarity with rotary encoders, I quickly recognized this problem as “noisy transitions” within the rotary encoder. In other words, it needed (more) debouncing. What I did was to install some capacitors to the signal pins and viola! it works almost as new. There is still a bit of debouncing problem but does not affect the responsiveness of the rotary encoder. If I experiment with different value capacitors, I would likely solve the problem, but for now this is good enough.

Other than this, the unit seems to be working properly. The only disadvantage is that now I cannot justify gutting it for parts 🙂

QUALITY CONSTRUCTION

From the golden era of Made in Japan audio electronics. Things are put together with more screws than seemingly necessary. Plus, this is the first device where I find the use of copper (or some copper allow) screws. The chassis is made of traditional stamped steel.

Nice brushed aluminum front panel (but the knobs are “metal looking” plastic)

The most ELNA capacitors in one place!

One of the last through-hole, hand-crafted audio components…

POWER SUPPLY

This receiver, old enough to be powered by a liner supply, is rated at 6x130W (780W for the amplifier section).

It uses a large EI transformer with a copper flux band. These bands are used In order to reduce the radiated flux from the transformer core,  acting as a shorted turn to the leakage flux (only), greatly reducing magnetic interference to adjacent equipment.

There are two 27,000 uF “Marantz” filter capacitors. Incredibly good looking! I believe they are made by ELNA (as every other capacitor is also ELNA). The heatsink behind the capactors is for the bridge rectifier. .

(Update: a reader alerted me that the caps are made by Nippon Chemicon. The logo is in plain sight)

There is space for two additional capacitors. A nice mod would be to add a couple of Panasonic 4-lead capacitors such as these: Panasonic T-HA 10,000-18,000 uF, 63V [link][link] (with care not to blow the fuse due to in-rush current during power-up)

The SR9200 uses 4 capacitors with higher voltage rating but lower capacity as shown in the photo below  [link]

ANALOG VOLUME CONTROL

The volume control is provided by two 6-channel Toshiba TC9482N volume control [link]

These devices control up to 8 analog channels (7.1 multichannel) that area available as pre-out but only 6 of them connect to power amplifiers

The input and outputs are buffered by NJM 2068DD opamps [link]. The “DD” grade devices exhibit lower noise specification. Where have we heard about these NJM2068?… From the development of the famous O2 headphone amp [link]

BOTTOM LINE: For those wanting to skip the Tech Section, the conclusions can be summed up as follows:

At gains less than 4X nothing overall could beat the $0.39 NJM2068 in the O2’s gain stage. This is especially true if you’re concerned about power consumption for battery operation.

…

Current prices of the 2068DD are $0.60 in quantity 1 orders [link]

DIGITAL TO ANALOG BOARD

Stereo D/A board uses CS4396 D/A (3 of them).

AMPLIFIER MODULE

A 6-channel module with forced cooling.

Local power supply bypass capacitors. Notice the space for larger size capacitors (the higher model SR9200 uses larger capacitors). Replacing these capacitors with larger ones (a 1000 uF nichicon KW [link] for example -maximum diameter is 16mm) would be an easy mod.

Local PS bypass capacitors in the SR9200

Output transistors: SANKEN A1492 (PNP) and C3856 (NPN)

DIRECT AUDIO PATH

There is an 8-channel analog input option (7.1 input) that bypasses all the digital processing. They are controlled by the analog volume chips and the output is available through the 8-channel pre-out. Six of those 8 channels are connected to the  6-channel power amplification module. This receiver can be used as a stereo tri-amp setup.

 

 

First One Build: Adjusting Operating Parameters

This should be the last info gathering post for building the First One Amp. Now I need to get a drill press in order to drill the holes on the heatsink…

The module is set to the correct operating parameters and tested at the factory. In case you need to check and readjust, most of the instructions can be found in this post [link] and following. The trimpots and test points are clearly maked on the board

 

OUTPUT BIAS CURRENT

The bias current when the amp is in idle (no input) is 220 mA for the output stage  [link] (200 mA absolute minimum if you have heat problems [link]), plus 60 mA for the rest of the circuit. This is a fixed value regardless of the supply voltage [link]. Trimpot TR3 is used to set the output bias current.Thus:

• Output bias current = 280 mA. (minimum 260 mA)

How to measure output bias current

The simplest way to measure the output bias current is with a digital voltmeter in current measurement mode. Connect the meter in series with the positive power supply wire. Alternatively you can connect a low value (1 ohm) resistor in series with the positive supply wire and measure the voltage across the power resistor. [link]

DC OFFSET AND VAS BIAS CURRENT

The DC offset and VAS bias are set as follows:

• DC offset = 0v (+/- 10 mv)
• VAS bias = 12 mA when cold or 15 mA when idle for 20 minutes

How to adjust DC offset and VAS bias

TR1 and TR2 sets DC offset and VAS bias current at the same time. Both works in pairs reciprocally as best explained by this post from the VAAS thread [link]:

• Adjust (both together) TR1 and TR2 clockwise: increase VAS bias current
• Adjust (both together) TR1 and TR2 counter-clockwise: decrease VAS bias

How much to turn TR1 and TR2 depends on the DC offset, so you must adjust both to arrive at the correct VAS bias and zero DC offset. Using two DMM would make the adjustment easier.

How to measure DC offset

• To measure the offset short the amp’s inputs and measure DC at the outputs.[link]

How to measure VAS bias

• Amp cold: 12 mA bias: measure 120 mV between TP1(+) and TP2(-) or between TP3-TP4 (doesn’t matter which pair of test points).
• Amp idle for 20 minutes: 15 mA bias: measure 150 mV between TP1(+) and TP2(-) or between TP3-TP4

PICTORIAL VIEW

Here is a diagram of the 3 parameters that can be adjusted:

• Use TR3 to adjust DC Bias to 280 mA
• Use TR1 and TR2 to adjust DC Offset to 0 V and VAS current by measuring voltage of 120 mV (cold) or 150 mV (warm)

First One Build: Ground Connection

LC recommends the following ground wiring (basically a ground lift) for the amplifier modules [link]:

Improved schematic for stereo connection in a single chassis. GND potential of each channel is lifted from chassis-earth potential, meaning connection is done via 1 k resistor and anti-parallel diodes. In this way EARTH potential interference currents are isolated from GND potential. At the same time GND potential of each channel also isolated in between.

The purpose of the ground lift device (the diode-resistor-diode) is a compromise between best sound and safety [link]

Since I don’t want GND to be complete floating I tied GND from both channels to EARTH potential via a DRD (diode-resistor-diode) chain. It is a compromise needed for a safety reasons.

(In fact…)

Best sound is at complete GND to chassis-earth isolation, so no DRD present. My demo amp, which I’m just listening at the moment, is in complete GND isolation (to EARTH).

The user is encouraged to install a switch that can short the ground lifting the device (the 1 k resistor and anti-parallel diodes) and experiment with both options [link]

You can even install GND lift switch on the back panel, shorting the DRD. So one position for GND (direct GND to EARTH connection) and another for GND lift connection (GND to DRD to EARTH connection)

So basically it is a “lifted ground” connection.

EARTH CONNECTION?

In the recommended hookup diagram above, the GND terminal of the Supply is connected to the GND terminal of the amp module. This is the obvious normal connection for proper operation. Each module has a single return path to the supply.

It is also necessary to connect the earth wire to the chassis for safety. This is to prevent exposing any harmful voltage in case a failure happens. Only if you have a “double insulated” chassis, then you can dispense with the earth connection (and this is what is called “Class II” appliance).

If one looks at the Hypex power supply specifications [link], they are built as safety Class II devices. This means they are already isolated with the minimum 6 mm from all possible conducting parts (its own metal frame). And can in theory be installed in a chassis without EARTH connection if you follow the double-isolation approach (meaning among other things that the wire you use for mains wiring has to be double insulated and having the expertise to double insulate everything else).

But in the normal approach of having an EARTH connected chassis, then the power supply’s metal frame becomes also connected to EARTH and with the 2-wire mains terminal, the power supply is in effect connected to the 3 mains wires in compliance with safety standards.

So from a safety point of view, signal GND connection to EARTH is not really a safety requirement (since the chassis is already connected to EARTH)

Now the question is “what is the purpose to connect the components GND terminal to EARTH?”

The answer might be in this application note from Hypex [link] where it says:

I can’t recommend separating the audio ground from the chassis ground, because that’s a recipe for making a radio receiver

So the reason is to prevent picking up electromagnetic radiation in the environment. And the best thing to prevent this is to connect the signal GND to the chassis EARTH.

The paper gives the following options:

• If you want to use RCA inputs, disconnect the mains earth and employ double insulated construction techniques.
• Use balanced (XLR) inputs. This allows the whole thing to be earthed unless the ancillary equipment has problems.
• Make a “pseudo-differential” RCA input. I still haven’t figured out whether or not I should post a detailed description of how to do this, because unless I manage to explain with perfect clarity it’s almost certain to generate large volumes of mail.
• Anything else (e.g. floating the amps inside a grounded chassis), but then you’re on your own if you hear your mobile through the speakers.

Thus connecting signal ground to EARTH it is about noise immunity.

Indeed, according to this article from RANE, signal GND must be connected to Chassis GND (which in our case, it is connected to EARTH ground) [link]

It is easy to confuse chassis ground and signal ground since they are usually connected together — either directly or through one of several passive schemes. The key to keeping an audio device immune from external noise sources is knowing where and how to connect signal ground to the chassis.

First let’s examine why they must be tied together… There are at least two reasons why one should connect signal ground and chassis ground together in a unit.

One reason is to decrease the effects of coupling electrostatic charge on the chassis and the internal circuitry. External noise sources can induce noise currents and electrostatic charge on a unit’s chassis. Noise currents induced into the cable shields also flow through the chassis — since the shields terminate (or should terminate) on the chassis. Since there is also coupling between the chassis and the internal circuitry, noise on the chassis can couple into the internal audio. This noise coupling can be minimized by connecting the signal ground to the chassis. This allows the entire grounding system to fluctuate with the noise, surprisingly providing a quiet system. Further coupling reduction is gained when the chassis is solidly bonded to a good earth ground — either through the line cord, through the rack rails or with an independent technical or protective ground conductor. This provides a non-audio return path for any externally induced noise.

The second reason to connect signal ground to chassis is the necessity to keep the signal grounds of two interconnected units at very nearly the same voltage potential. Doing so prevents the loss of system dynamic range where the incoming peak voltage levels exceed the power supply rails of the receiving unit.

WHY USE GROUND-LIFT?

According to this document on audio grounding [link]

Some people believe that it is necessary to isolate the system star ground from the chassis and safety ground in order to have a hum‐free audio system. However, if all of the components in the system have their grounding implemented properly, there is absolutely no need for ground isolation,

Although isolating the grounds may eliminate a ground loop, it does come with two penalties:

• First, since the signal reference (signal GND) is not directly connected to the chassis, the chassis is not an effective shield for the electronics
• Second, since the power common is isolated from the safety ground and connected to the signal reference, any AC leakage current from the power supply may flow through the signal reference to get to the safety ground in another component.

If you must isolate the grounds; never, ever, for any reason, disconnect a safety ground (chassis connection to EARTH ground) or fail to provide a safety ground in any equipment that you build. First, it is unsafe and second, there are equally effective methods of isolating grounds that do not come with the safety hazard. The following figure shows two such methods.

First is to provide a “ground lift” switch between the two grounds to be isolated…

A better solution is to provide a Safety Loop Breaker Circuit (SLB). This circuit will allow the current from a fault to flow to the chassis and also provide ground isolation under normal, non‐fault conditions.

You can find a circuit for a ground loop breaking (or SLB) from Elliot Sound Products [link] which is in principle similar to what LC is proposing. Thus the recommended circuit provided by LC is the proper method to avoid ground loops and be able to interface with upstream components with less than ideal ground implementations.

In my diy builds I’ve never connected the signal GND to EARTH, but have always connected EARTH to chassis. I think it would be a good idea to try connecting the signal GND to EARTH with a break switch to compare.

DOES GROUND-LIFT COMPROMISES SAFETY?

Based on the discussion above, ground-lift (that is not connecting signal GND to EARTH) does not seem to compromise safety. The fact that you can purchase double-isolated appliance with a two-wire power plug also says that it is not required to have a current path to EARTH in case of a fault. But don’t take my word for it. I am not a safety expert, just using a little bit of common sense. In my projects, I never connect the signal ground to earth ground but ALWAYS connect EARTH to the chassis and when connecting EARTH to chassis is not possible (like using a wood plate) then I make sure there is plenty of air gap between the component and my fingers…

First One Build: Power Supply Selection

LC tested three First One amplifiers, each one equipped with different power supply [link]:

• Hypex SMPS1200A400 [link]
• AudioPower DPS-500/63
• Connex SMPS2000R [link]

AB comparison 1: Single Connex vs Dual AudioPower

It was not really hard to recognize the better PS, since bass was really on a weak side of Connex, probably the reason lies in the use of a single SMPS2000R for both amp’s channels, nevertheless the sound-stage was narrow, instruments weakly presented, thin bodies, no real feeling of involvement into the music. DPS-500/63 were in dual mono configuration, meaning one per channel. Transparency and details resolution was very similar but the bass was one step ahead over the Connex. Still no real musical involvement presentation here, sound more or less stuck to the speakers, instruments with weak bodies, sound-stage not well formed.

Since both were regulated SMPSs we got an impression that the problem lies in the voltage regulation principle, too much interfering in audio signal is probably the worst thing it could happen from amp’s power supply.

It is interesting to point out that without AB comparison, LC had this to say about the Connex supply:

Cristi, your second SMPS2000R is just WOOOW, driving both channels. This is the one packed with Rubycons on primary and secondary side, set to +/-63 V, no hum, dead quiet silence, rock stable imaging, resolution is at the top notch, no sign of sibilants on vocals, liquid like sound, bass is simply awesome, like a day and night different from first one tried, also much better than SMPS1200 [link]

Yes, a day and night between the two SMPS2000R, named them A and B versions. As I can see the only difference is in secondary cap bank, maybe also something else, surely Cristi would have something more to say about them. Anyway version B with Rubycons is an absolute winner up to now, doubtless.

SMPS1200 provided better bass than Cristi’s A version, but that’s about it, both Connex gives higher headroom, also more power because of higher and regulated rail’s potential, at the end resulting in more stable sound imaging. [link]

AB comparison 2: Single Hypex vs Dual AudioPower

Here is the latest version of the Hypex supply. The entire PCB was updated in 2013 according to the data sheet [link]

After half an hour of mental pause, test continued between (single) Hypex SMPS1200A400 and (dual) Audio Power DSP-500/63. First we listened to DPS-500/63, musical impression stayed very much the same as it was in the first session. Weak presentation left us more or less cold, music is simply too thin, uninvolving.

Hypex’s turn then settled the things where they should belong, suddenly real music in the room. It was immediately clear to all of us who’s having magic stick in its sleeve. SMPS1200A400 shocked us with music so real that the other two seemed like that there was something broken in them.

Hypex presented bass so low and strong that our jaws just dropped for a while, dense wall of air moved with lowest possible frequencies we could still hear, still with ease. Sound-stage completely another story from previous contenders, no speakers in the room only musicians and instruments, this time with fully developed bodies, atmosphere of a recording stage, whether real or artificial, presented in full scale. SMPS1200A400 puts the First One amplifier in the first league of power amplifiers no matter the price level.

AB comparison 3: Single vs Dual SMPS1200A400 [link]

The result is not so far from our expectations, tight similarity with very slight differences noticeable on momentary A-B test; otherwise, on not so closely conducted comparisons, it would be very hard or even impossible to distinguish the two configurations.

In which cases to choose single or double SMPS solution greatly depends on the speaker’s efficiency and impedance, these two parameters dictate how power hungry your speakers are and of course how loud you want them to be.

The ultimate and preferred First One solution is still dual mono configuration, although the use of a single SMPS completely fulfills the needs in a smaller system.

AB comparison 4: Linear vs AudioPower [link] (Comparison on-going…)

Conclusion

Well, after the not-so-subtle remarks in the AB comparison performed by LC, it is almost impossible to argue the fact that using anything else than a Hypex SMPS1200A400 supply would “rob” performance out of the amplifier modules. A plus is that the amp modules are factory calibrated and tested with this supply in mind. In addition, especially for a budget-constrained build, the AB comparisons also showed that unless there is a need for high current demand, a single SMPS1200A400 would sufficiently fulfill the designed performance of the First One amp. Further, the Hypex SMPS1200A400 is competitively priced against the other two offerings.

Best choice

If one wishes to confirm the results or adjust certain variables in the comparisons, it would be difficult for the diyer to replicate even some of the tests reported by LC. First several amps were configured with the different supplies for a immediate AB comparison. Second, one would have to procure the different supplies for the test. This would be cost prohibitive for the common diyer. In addition Hypex has a long and excellent reputation for high end audio not just a a provider of high fidelity products but also as a technical innovator, and thus there is very little chance to find a superior supply for this application.

Why Hypex supply outperforms Connex supply?

Having examined the Connex supplies in detail, I find that the use of soft switching approach minimizes the generation of EMI while increasing efficiency. The hypex datasheet does not say whether it is a “soft” or “hard” switching approach. It merely says “The SMPS1200 is optimized from the first phase of design to final implementation to realize the lowest possible EMI signature required of the most demanding audio applications” this could very well mean the use of soft-switching and/or more aggressive output filtering. So from a SMPS switching approach point of view, one cannot say why one would sound better than the other.

Therefore, the difference in audio quality seems to be fully attributed to the unregulated output nature of the Hypex supply. This comes as a surprise because the audio implementations have been moving toward regulation and now we find that SMPS with regulated output seems detrimental to audio quality at least in this instance.

LC believes that regulation “interferes” with the audio signal [link]

For those interested in more technical details, the Connex supplies were tested extensively [link]. And even thought they exhibit superb performance and I remain a fan of Connex supplies [link], I cannot justify using them with the First One Amp modules in light of the comparison presented here.

AudioPower develops Unregulated PS

It is worthwhile to note that AudioPower has recently developed Unregulated versions of their Audio SMPS, perhaps a testament that “unregulation” has sonic advantages (or just competing with Hypex). [link]. I’ll have to admit, they are best looking.

My initial choice

For now I will use the power supply of the Adcom GFA-5300 AM. It generates +/- 52 volts and according to spec, can supply a max of 720VA

First One Build: Heatsink Selection

One of the most important parameters for proper operation of the Fist One amplifier module is adequate heat dissipation through a large enough heatsink. The amp module operates in class AB with an idle current of 280 mA.

The idle dissipation of the First One is >30 W at +/-63 V, plus audio power dissipation easily adding extra 50-70 W, so 100 W all together to dissipate.

To calculate the temperature of the heatsink during operation: 100 W * 0.5 K/W=50 K added to room temperature (25 C) resulting in 75 C heatsink temperature. At that point silicon die in output transistor is around 100 C and that is somehow at max acceptability. To calculate idle temperature: 30W* 0,5 K/W=15 K (or C) added to room temperature resulting in 40 C. [link]

Supply DC current to the First One module without input signal present (idle current) is 280 mA, multiplying it with 120 V rails potential, gives 33.6 W of total quiescent power dissipation per module, so in stereo total 67,2 W. That is serious thermal loading for the chassis and heatsink if one would want amp to be below 45 degrees in a room environment. [link]

LC recommends “any heatsink having 0,5 K/W or even lower”. Something like Fischer Elektronik FK157 [link]. Below are the heatsink profiles of the FK157 and other similar profiles that will yield 0.5 K/W dissipation or better. These were extracted from the Fischer Elektronik catalog [link]

Notice that by comparing the 3 profiles shown above, in order to achieve a 0.5 K/W dissipation you would need:

• 2″ of SK501
• 2″ of SK 586
• 2″ of SK 157

Seems longer fins only help if you need a dissipation factor lower than 0.5 K/W or even lower than 0.3K/W

Even a shorter profile would yield a dissipation rate of 0.5 K/W. In this example a 4″ heatsink would achieve a dissipation rate of 0.5K/W

The Semelab application note has a extensive section on heatsink selection [link]. If you read the whole thing, basically the bottom-line thing to do is that the heatsink shall not exceed 70C during operation.

Factory Chassis

LC provides appropriate heatsinks as part of the factory chassis (which cost Euro 300 plus shipping). Following are the photos of the “factory” heatsinks ( I think they are SK 157 with height 70 mm, so having a dissipation coefficient of a bit less than 0.4K/W). The chassis is beautiful and built like a tank. If you want the best, this is it.

In order to obtain the factory heatsinks, you need to purchase the chassis (300 euros plus shipping – I would think US$60-$100 for shipping based on eBay examples. So total cost would be US$450-$500) [link]:

Although it is highly desirable to have an enclosure that is built at the same high standards as the amp module, if budget does not permit, there are other options.

Chinese chassis from eBay

The is the the lowest cost for a chassis plus heatsink meeting the required dissipation rate [link]. This case costs about US$ 160 including shipping.

The heatsink size for this case is 300x50x67mm with a profile similar to SK501 but with the fins 10 mm longer. At 67 mm height, Likely it exceeds the required 0.5 K/W dissipation rate. It probably rates at 0.45K/W. (This is just theory in practice you may need a larger (taller) heatsink depending on different factors such as ambient temp, etc)

You can find an example implementation of this case with the VSSA amp here [link].

A similar but with taller heatsinks can be found here [link] and here [link]

Heatsink Only

If you want the minimum cost and If you live in the USA, a good source is “HeatsinkUSA”. High quality and good prices. The largest one seems to fit the bill [link]. Specs are:

• Width is 10.080″
• Fin Height is 2.5″
• Base Height is .375″
• Weight is approximately .99 lbs per inch
• C/W/3″: approximately .80 (for a 3″ heatsink)

This heatsink is similar in profile to  SK524 above except it has one less fin but the fins are much larger at 50 mm. If we use the dissipation curve of the SK524 we find that a 4″ heatsink will meet the required 0.5K/W dissipation. Note that the published thermal dissipation specification for this heatsink is 0.8 s for a 3″. If we go by the dissipation curves shown in the Fischer Elektronik catalog, then this values seems too conservative. But in order to be safe, a 5″ heatsink would likely be more than sufficient.  A pair of 5″ heatsinks would set you back about $90 including shipping.

Thrift Store Amp

Even cheaper than getting heatsink is using an old amp from a thrift store. If you are lucky, you may find an old amp with large heatsinks. I had purchased a used Adcom GFA-5300 amplifier from the local thrift store for $15. This was a few years back. nowadays, even thrift stores are drastically increasing the price of used audio equipment. I would say this amp would probably sell for $50 if bought today.

The heatsink of the Adcom has the following dimensions:

Width: 200 mm; height: 90 mm; depth (fins): 55 mm; base plate thickness: 5 mm; number of fins: 20.

The closest profile I could find from the Fischer Elektronik catalog is the following:

As can be seen, the Adcom heatsink is a bit wider, the fins a tad longer and it has 4 more fins. I would say at 90 mm in height, it would easily meet 0.6K/W. but it does not meet the minimum requirement of 0.5K/W.

Using the power supply of the Adcom Amplifier

What if we use the power supply of the Adcom which provides +/- 52V? We can calculate the required heatsink dissipation with this supply by following the example given at the beginning of this section and the following requirement [link]:

As we don’t want to have more than 45 C in idle, please use heatsink having thermal coeficient of 0,5K/W or less for each channel.

First One module has 35 W idle power dissipation when supplied from +/-63 V PSU.

The idle current of the Amp is 280 mA (how to measure [link]). Even at a lower supply voltage this bias requirement is fixed [link]

Thus at +/- 52V supply we get 29 Watts. With a heastsink of 0.6K/W we get 29*0.6=17.4 C. Adding the room temperature of 25C we get 42.4K which is within spec but this is only at idle.

The service manual of the Adcom Amp gives this power data:

I plan to use this Adcom amp for my first build. It seems to have adequate heatsinks. I will have to build up the amp to know for sure.

Summary of choices

• Factory case: ~$450-$500
• eBay Chinese case: ~$150-$200
• Heatsink Only: ~$90
• Thrift Store Amp with large heatsinks: ~$30-$60

First One Amplifier Module

The First One Amp module [link] is a high performance (High Fidelity?) and yet very affordable class AB current-feedback amplifier module.  It establishes a benchmark for price/performance.

Developed by “Lazy Cat” (LC) at diyaudio, it is the big-brother commercial version of the DIY VSSA (“Very Simple Symmetric Amplifier) [link] and incorporates all the knowledge obtained from that project. Whereas the VSSA was fully open and fully diy, the First One amplifier is available as a factory built and tested module. Available for diyers as well as OEM to manufacturers, the module has been seen in a finished amplifier for a road show in Slovenia [link].

Photo of First One’s little brother: completed VSSA module (built by LC):

FIRST ONE MODULE

Since I am new to this module and was not aware of the VSSA, I’ll use this and following post to gather my knowledge for my amplifier build. The information is mostly from the diyaudio threads, but there it  is spread out all over the place and hard to find.

Photos of the First One Amplifier Module.

Use of name-brand “audio grade” components…

The current version is V1.2. There is a V1.3 that has been developed but not quite yet available for sale. For those of us with V1.2, LC has promised to send modding instructions but only to those that have completed the build of the amp.

Thermal coupling for these two transistors. The schematic is not public since this is a commercial product.

Notice the adjustment pots (TRx) and the measuring points (TPx)

Output Power transistors are Semelab “ALFET” double die MOSFET N and P-channel pair, rated at 250 W and 16 Amp continuous current. These are specially designed for audio applications [link]:

• The N-channel device is: ALF16N16W/ALF16N20W [link]
• The P-channel device is: ALF16P16W/ALF16P20W [link]

SPECIFICATIONS

Module Size

100 x 50 x 40 mm (W x D x H). [link]

Supply Voltage

+/-40 V to +/-63 V

DC coupled

The amp modules are DC coupled, no capacitor in front of the input stage.

Measurements

I’ll compare the specifications of the First One module (FO) [link] vs an old Hitachi amp [link] and an Adcom amp:

Parameter	First One	Hitachi HMA-7500	Adcom GFA-5300
Max Power 8 Ohm	150 Watt 0.05 THD	80 Watt 0.005 THD	80 Watt 0.018 THD
Max Power 4 Ohm	230 Watt 0.05 THD	80 Watt 0.005 THD	125 Watt 0.018 THD
Bandwidth	3 Hz to 3 MHz (-3dB)	5 Hz to 100 KHz (-1dB)	3 Hz to 130KHz (-3dB)
THD	0.0034% (100 Watt)	<0.005% (80 Watt)	0.02% (125 Watt, 1KHz)
IMD	0.003%	<0.008%	<0.07%
SNR	110 dB	118 dB	>100 dB
Input Impedance	10 Kohm	47 Kohm	50 Kohm
Damping Factor	>2000 (4 ohm)	100 (8 ohm, 1KHz)	>350
Year Introduction	2014	1980	1995

According to published specifications, the First One amp has very impressive specifications and overall best of the bunch. The old Hitachi has still has very impressive specifications (but at a much lower max power).

Damping factor

Measurements performed in order to determine Zout and consequently the damping factor (DF). A sinusoidal signal of 100 Wrms at 20 Hz, 1 kHz and 20 kHz was passed onto a 4.08 Ohm load resistor, measured with FLUKE 289 True RMS Multimeter and here are the results. [link]

20 Hz, 100 Wrms/4.08 Ohm:

• DF(20 Hz)=Rload/Zout=4.08 Ohm/0.00121 Ohm=3372

1 kHz, 100 Wrms/4.08 Ohm:

• DF(1 kHz)=Rload/Zout=4,08 Ohm/0,0004 Ohm=10200

20 kHz, 100 Wrms/4,08 Ohm:

• DF(20 kHz)=Rload/Zout=4,08 Ohm/0,00162 Ohm=2519

Very large damping factor by itself likely means that the amp itself would not be the limiting factor for controlling the oscillations in the speaker. This means that other factors (such as speaker cable impedance) would contribute more to the damping factor seen by the speaker. The speaker’s own impedance is the mayor contributor…

A PERFECT MATCH WITH R2R DAC?

Seems a perfect match for the upcoming discrete R2R DAC. The amp being single-ended (and DC-coupled) can take the output signal straight out of the resistor ladder. In addition, being wide-band would further benefit from R2R conversion (as opposed to delta-sigma) because the R2R DAC does not generate high frequency noise.

Hypex UCD-180HG and Connex SMPS300RE

(See update to listening impression)

Finally finished updating the power supply for my Hypex UCD amps. Each amp module was upgraded from a linear, unregulated power supply to a switched, regulated power supply from Connexelectronic. Even with all the debate between linear vs switching supplies, I think the power switching technology (and affordability of that technology) has finally caught up to the advantages of linear supplies (low noise) and surpasses them by providing regulation. Not to mention weight, size and efficiency. There is also no excuse not to try because Connex sells these modules under $60. Just the transformer for a linear supply of equivalent power would cost you more than that.

ORIGINAL CONFIGURATION

This is the original configuration for the UCD 180HG amp modules: ~40V unregulated linear supply, using a 150VA toroid transformer from Apex Jr, and a snubberized PS module from Chipamp. At that time, these cost me about $60 for each which coincidentally is the same price as the connex unit.

This configuration has served me well for more than 5 years. It is a bullet-proof design, and have no complains about it.

THE NEW POWER SUPPLY MODULE

The new configuration uses a +/- 45v Resonant SMPS with regulated outputs, the 300RE. This is based on the latest resonant SMPS designs resulting in lower noise and higher efficiencies. I’ve described this PS in more detail in this post [link].

THE COMPLETED AMP

The Hypex modules are “far away” from the power supplies to avoid any interference. I’ve also installed a metal plate in between for further isolation.

Since the recycled case had too many holes in its face, I installed individual power switches for each of the power supply modules and also enable switches for each of the amp modules. These switches were already present in the old configuration so I just imported them here. In addition to allowing individual control of the amp modules and PS modules, these switches were useful while I was building up the amp for testing and safety.

The PS modules have been set for exactly +/- 45V operation which is the specified recommended “typical” operating voltage for the amp modules. The output voltage can be adjusted approx +/- 10%.

I followed the hookup configuration of the data sheet, which includes fuses for the positive and negative DC power lines for protection of the amps and the speakers.

Chinese binding pots from eBay… Pretty good looking and they are “fat” making them very easy to use.

The PS modules each incorporate input (and output) EMI filters, thus no additional filtering is necessary. However, I added a cable ferrite which seems to be “standard practice” for switching supplies.

I use the AUX supply just to power an LED :-).

The Hypex UCD180HG modules.

These were the original version of the “HG” models. I believe the current version has a revised PCB and the maximum voltage also increased. According to the datasheet of this original model [UcD180HG_datasheet2] the operating voltage range is as follows:

45V is likely the optimum PS supply voltage as the performance data is based on this value.

Input signals are carried by trusted Cat5E patch cables. I also have a passive RC filter currently configured with a -3db cutoff at approx 75KHz. The reason for this filter is because I use the Buffalo II DAC in voltage-out mode connecting straight to the Amp. The values are R=220 ohm, C=10 nF

I may adjust the filter to the values used by the Legato I/V which I believe has a cut off frequency of 400KHz.

Doesn’t look too bad even…

Got to find something to put in those 3 empty holes 🙂

HOW DOES IT SOUND?

It was imperative that I finish the amps for proper listening with the Amanero USB interface (which quickly made playing DSD files from a computer an affordable reality). My queue of projects (Ian’s FIFO bard, the new AVCC supply for the Buffalo DAC) also requires a proper working listening environment. Previously I had disassembled one amp in preparation of the power supply update and for a while had been “listening” to different experiments with just one channel.

Anticipating the sound of the “new amp”, my own experience tells me that changes in the audio path are often subtle and big differences are far in between  and in general, I also think not having inflated expectations and not rushing in trying the latest and greatest allow me to be more objective in perceiving any changes. Thus for this power supply upgrade, I did not expect any big changes in the sound.

After only one listening session, I was pleasantly surprised!

Upon powering the amp and playing the first track, I immediately noticed the huge “surrounding” soundstage. Hmmm, this is weird. OK, fixed that. I had switched the channels :-). Play again…

This time I noticed the soundstage again. From memory, I did not recall ever experiencing the soundstage this way (maybe this is somewhat related to having listening to one channel for a while but…). Certainly it was not this expansive, not this large, not this 3D. Yes the speakers disappeared, but now the speakers disappeared even more:-). Pinpointing my attention to the speakers, no sound was coming out at that location at all. This sounds really, really good. Play more tracks…

The bass was also different. Now it was more controlled. I played and compared tracks where from memory I thought the bass was a bit booming. Now they sounded more controlled, more defined. The “boominess” was gone.

The rest of the audio spectrum seemed the same as before, perhaps a tad more “crispiness” in the higher frequencies, but can’t tell for sure.

Overall, music appeared more dynamic. Perhaps the additional headroom (45V supply vs the old 40V supply) combined with a better bass definition creates that perception. The new PS can also deliver more power.

In summary, this has been one of the best upgrades I’ve had in my system. This is another indication that the quality the power you use is perhaps the most important ingredient for a good sounding system. Certainly it makes the obsession of tweaking of the power supply almost justifiable:-). It is the first time I have use a regulated power supply for an amplifier and it is certainly consistent with the trend in the audiophile community of migrating from linear-unregulated supplies to switching-regulated supplies. Hypex has purposely designed a switching supply for their latest NCore modules.

The Connexelectronic PS modules are not only very affordable, but incorporate the latest low noise SMPS soft-switched resonant technologies. In addition, the 300RE incorporates more filtering in the form of an output CLC network. If you have UCD180 amp modules or equivalent, this upgrade is highly recommended.

Update (12/15/12): False alarm!

The larger soundstage was caused by inverting the balanced connection for one of the channels. After listening for a while, I noticed that the soundstage was somewhat artificial. So I traced the speaker cables, the connections from the amp module to the speaker and finally the input to the amp modules. One of the input wires was reversed.

After fixing this error, the soundstage “collapsed”, but now more realistic. Now the soundstage is more or less the same as before. At the least this showed me how phase issues would sound like.

The bass did remain more controlled, but not as pronounced (not as tight?)  as before. Overall, still an improvement to the old PS and still highly recommended.

I did play a “silent” track to see if there is any noise coming from the amp: total silence. Increasing the volume to zero dB did show a faint hiss if one would press the ear to the speaker. But the hiss was coming from the DAC! The amp is totally silent.

FURTHER READING

Here is a similar project using the SMPS500 with the UCD400: http://diyclassd.blogspot.fr/

Impressions shared in diyaudio:

[link] I exchanged in my power amp with two Hypex UCD400HG modules the two 1KW toroid transformers and the New Class D PSU with split foil caps to two SMPS500R.

What is the difference, in my opinion not so much. Maybe the bass has a little bit more pressure but it is only a feeling. For the rest everything is the same like before.

For that reason the SMPS500R stays in. These things are absolute quiet and the best of all they save energy without any compromise in sound quality.

A SMPS300 + UCD180 Kit (illustrated build guide) [link]