Audio Research VS155

[Michael Maurice]

I've got this amplifier on my bench.

I dont have a circuit for it but the output valves are 6550A's

I've found 2 of the 100 ohm resistors that feed the G2's have burnt out, the HT is high because the semiconductors Q1-Q5 have failed. However the part numbers are special to Acoustic Research and I've no idea what they are.

The -75V for the grids is there but at around -85V

Presumably one or more of the o/p valves has had an internal short circuit..

Is there anything else I should be looking for? Does anyone know what I could use for replacement semiconductors?

[Diabolical Artificer]

Service manual here - https://www.ultimateservicemanuals.c...390-pdf-on-dvd - purportedly, though for cash. These amps aren't a job for the faint hearted, some of the circuitry I've seen in their amps is complex, using DC servo's, gyrator's and other uncommon circuit's and to make matters even trickier the Q's aren't ID'd as you say. They like using a lot of FET's and AFAIK finding the part #'s is impossible AR guard their knowledge like Rottweilers.

I found a circuit for the VS55, but the front end is solid state, but is does have the circuit for the regulated PSU, they use pretty similar circuitry in a lot of their amps, with more time to search online you might find something nearer the VS155.

Good luck with this one, but at six grands worth of amp I'd be wary. Andy.

[Craig Sawyers]

Do you mean VSi55 - an integrated pre/power?

Audio Research are (or certainly were when I owned a pre and power amp) pretty good at supplying schematics.

The problem as you've found is that they grind the part number off and use up to 3 colour code dots. Cryptic, but I think that is because their circuits are rather critical for Vp and Idss of FETS and beta of transistors. So they grade them for different circuit applications. It prevents repairers, tinkerers or those who want to try "improved" devices from messing around.

I'm sure they will also supply replacement transistors, but they are unlikely to be cheap.

You've been unlucky to come across a catastrophically failed unit, because they are usually pretty bomb proof and often contain a shed load of protection circuitry.

Craig

[Craig Sawyers]

The VSi55 schematics and parts list is on Elektrotanya, attached

Craig

[Craig Sawyers]

Also found a comment

"CHANGE OF OUTPUT TUBES (2007)
VSi55 had its output tubes changed from 6550EH (Electro Harmonix) to 6550C (Svetlana) at the production line. The output tube bias was changed to 55mV, providing better sonics and improved reliability."

From https://www.arcdb.ws/

[GrimJosef]

Originally Posted by Craig Sawyers

... The problem as you've found is that they grind the part number off and use up to 3 colour code dots. Cryptic, but I think that is because their circuits are rather critical for Vp and Idss of FETS and beta of transistors. So they grade them for different circuit applications ...

This mirrors my experience with the only bit of AR kit I've ever worked on. It used digital volume controls and to ensure tighter matching between the L and R channels AR built a selection jig and used it to group the chips into colour-coded sub-groups. They would sell you a replacement set all of the same colour. But you'd better be sitting down when they told you the price. By the time the item came to me the chips were obsolete anyway. I managed to find a source in Germany where prices were low enough for me to order more than I needed. I then built my own selection jig and put together a set which would do the job. Once the kit was working again I have to say it sounded great and measured brilliantly.

Cheers,

GJ

[Radio Wrangler]

Interesting to look at.

They DO like their FETs.

Looking at the HV power supply, one FET is used as a constant current source "Q2" it will have to be selected because of the device to device spread in order to get the current within reasonable bounds.

Q4 and Q5 are a differential amplifier/long tail pair and these will need to be a reasonable match because any discrepancy will be magnified and affect the output voltage. This power supply relies heavily on tighter spec devices than you can buy. The manufacturer must, as said, have an extensive parts selection process. I wonder where the rejects go? Also the danger with such selection processes is what if the next several batches of new devices that come in are all rejected and they start to run dry on acceptable parts? There are no guarantees on yields. What benefit does this difficulty provide? Well they get a lot of gain out of only a few devices. It could certainly have been done better from a point of same performance from a circuit that is both manufacturable and repairable. High voltage linear regulated power supplies are as people hereabouts know, a bit prone to failure.

There is a lot of stored energy in the reservoir banks and if there is a momentary durrent demand pulse, the big FET IDss isn't going to be much help, the voltage drop across that 100 Ohm series resistor is going to upset Q2. There is no current limiting mechanism of any predictability.

The amplifier proper doesn't look too terrifying. Q1 is another FET set up as a constant current source, and will need selecting to get the bias current of the long tail triode pair V2/V3 right, getting the quiescent anode voltages right.

+354v supply and +225 on the anodes gives 2.58mA down each of the 50k anode resistors, so 5.16mA through the constant current FET.

You can work out what you're selecting for and build test rigs. You can look at voltages and currents and choose devices with adequate ratings. It's reverse engineering, but at some point you begin to notice you're doing almost as much work as the original designer had to do.

It's got overall feedback from the output winding of the output transformer which is bog standard, but it also uses that winding gor feedback to the cathodes of the output pair. A neat variant on the Quad II.

There is however, only 1 Ohm current measuring resistance in the cathodes. This amp is utterly reliant on someone setting up quiescent currents by measuring the drops on those 1 Ohm resistors and setting the bias pots right. It is then utterly reliant on the valves staying stable. Oh, an open circuit pot slider-track contact would be disastrous.

Burnt G2 stoppers suggest either a failed valve, or a burst of oscillation.

So, it's probably OK when it's working. A pig to manufacture, and a double-pig to repair.

It contains a few boutique-y design choices which cause this. D minus for engineering, could try harder.

David

[GrimJosef]

Originally Posted by Radio Wrangler

... The manufacturer must, as said, have an extensive parts selection process. I wonder where the rejects go? Also the danger with such selection processes is what if the next several batches of new devices that come in are all rejected and they start to run dry on acceptable parts? There are no guarantees on yields ...

Very high-end audio manufacture is different from most other businesses (maybe very specialised defence gear is similar ?). Production volume is relatively small. The pre-amp that I worked on sold, I heard, a total of 12 units in the UK and that was more than in any other country, including the US where AR are based. Its retail list price was £24,000.

For that you can afford to buy ten times the chip stock at the start of production that you are ever going to need, and throw eighty percent of them away. Keep a small box full for spares and when they're gone, they're gone. Explain to your disappointed customer that this is how it is with ultra-special parts, and they wouldn't have wanted anything less than the most special would they ? Never mind. We have a new, and even better, model for sale now.

Cheers,

GJ

[Radio Wrangler]

Originally Posted by GrimJosef

The pre-amp that I worked on sold, I heard, a total of 12 units in the UK and that was more than in any other country, including the US where AR are based. Its retail list price was £24,000.

Oh dear. A much better high voltage regulator which regulated every bit as well and tolerated device tolerances would not have been difficult to design. Also the current source in the amplifier could have been done with properly tolearenced circuitry and made to work better, too.

An example is the ring-of-two reference published in Wireless World, but at such low currents a resistor from the HT supply into a current mirror would avoid the minimum currents zeners require AND it would make the current proportional to the HT voltage, which with the anode resistors in the valves would make the anode sit at a tracking fraction of HT, thus optimising headroom (Neat, or what?)

I don't have the stomach for doing boutique designs. They seem to have difficulties and the difficulties are avoidable without disadvantaging anything else. Either someone didn't know of better ways, or it's fakery. Without all the selection and its wasted parts and effort, maybe the preamps could have sold for a fraction of the price and sold disproportionately more and worked better. Net results: More profit for the manufacturer, more people with products which make them happy, and as a designer, I'd rather live in a world with more happy people in it. I don't get a kick out of exclusivity. If everyone else's amplifier sounds better, it doesn't make mine sound worse.

No, I think they could have tried harder. D minus.

David

I my (and many others) opinion anything that need selection of devices or won't work quite well with components +/- 20%* isn't a good design (given offsets etc.) They must have a FETish, sorry, couldn't resist that.

I have seen LED series resistors specified to 1% to get 10mA through the LED, totally ignoring a) the temperature coefficient of the Vf of the LED, b) the variability of the LED brightness vs. current, and c) that the eye wouldn't notice anyway.

*I would go as far as 50% for at least working but not so well.

[Radio Wrangler]

Mea culpa on 1% LED series resistors, but only because we'd standardised on 1% 1/8W resistors to simplify the numbers of types of components in our products.

David

[GrimJosef]

Originally Posted by Radio Wrangler

... Without all the selection and its wasted parts and effort, maybe the preamps could have sold for a fraction of the price and sold disproportionately more and worked better. Net results: More profit for the manufacturer, more people with products which make them happy ...

These really aren't consumer products in the same sense that white goods or large-screen tellies are. There is 'price sensitivity' in both markets, but it's a very different (in fact practically opposite) sort of price sensitivity. Buyers of new AR stuff want the high price and exclusivity. AR would like to sell more units of course, but they realise that dropping the price significantly would, to begin with at least, allow more room for their competitors to claim 'ours is more expensive, so must be better'. Heading downwards in price would probably hurt them for the first few tens of percent.

On average the selling price of hi-fi seems to split one third to the retailer, two thirds to the manufacturer (in high-end there are really no middlemen). With cheaper products the parts and assembly cost are significant - maybe more than 50% of the factory gate price. With high-end stuff there's little point in cost-paring in these areas. They don't make up a large part of that price. The manufacturer's overheads (product development, publicity, simply 'running the business') consume most of it, because the sales volume isn't large. AR aren't competing with LG or Panasonic.

I should say the pre-amp that I saw worked as nearly perfectly as any I have ever measured. It couldn't have worked much better. The spec on the digital volume control was that the L and R channels would be matched at every one of 104 steps to within a small fraction of a dB (0.2dB IIRC). It was better than the chips' spec which, for their day, was quite good. So the selection really was necessary (whether the 0.2dB matching was is debatable). The noise and distortion figures were a long way better than anything audible.

That said, the cause of the fault was a design issue, but not really an electronic one. The unit was pcb-based and the valves and the semiconductors shared the same large double-sided board. The valves need ventilation. So there are slots in the top of the case. It's only a matter of time before something falls through one of the slots. And because there are tracks on the top surface of the board the 'something' can short the valves' circuitry into the semiconductors'. As sure as night follows day, mixing valves and semiconductors results in blown semiconductors.

Cheers,

GJ

[Michael Maurice]

I have got the diagrams, I forgot to post them up so here they are.

I'm very tempted to tell the owner that I cant get parts. but I think I will try the importer which is Absolute Sounds in London.

I'm wary of fitting expensive parts to this amplifier only for them to blow up again.

[Radio Wrangler]

The "1016" MOSFETs have to be depletion mode parts, and selected for Vgs at 5mA Ids
I've no clues to what types they would have been before selection. At a guess something like Supertex/Microchip DN3545 DMOS depletion mode FET 450v but the pinch-off gate voltage could be anywhere between -1.5 and 4.5v Now with 338 ohms in the source lead and a current much closer to pinch-off than to full on... 5mA times 338 Ohms is 1.69v so I think they're selecting them for the lower end of the Vp spread. This is just a guessed part and then a wet-finger calculation to illustrate what they've been up to.

Mouser want 53p each in singles for these, but as you'd need them to the bottom end of their Vp spread, who knows how many you'd have to buy to get lucky? Parameters are strongly influenced by batch. You may have to buy enough to get different batches and still cross your fingers.

GJ has explained their rationale. But in my view it is grossly incompetent engineering to do it this way. The circuit could have been made tolerant of part tolerances and they would still have been free to charge impressive/reassuring/extortionate prices for it.

I think you're stuck buying their selected parts. It would be interesting and see how much they value them at. Of course, you don't know how deeply you have to go replacing things to be sure there is no conflagration at switch on.

What would I do? It depends. I could take some guesses at likely parts and do some selection, hoping to find some suitable, or I might indulge in a bit of redesign. Of course, this would either be seen as blowing away all the fairy dust, or it could be seen as a wonder mod increasing the fairy dust, if I can find a pointy hat with stars on it and the word "Rincewind". As paying commercial jobs, I'd steer clear of them.

David

[bikerhifinut]

The word's "WIZZARD" David, but you know that as a now out of the closet Terry Pratchett afficionado!
Sorry.

A.

[GrimJosef]

Originally Posted by Michael Maurice

I have got the diagrams, I forgot to post them up so here they are.

I'm very tempted to tell the owner that I cant get parts. but I think I will try the importer which is Absolute Sounds in London.

I'm wary of fitting expensive parts to this amplifier only for them to blow up again.

Absolute Sounds were, last time I checked, AR's accredited service agent in the UK, as well as being the importer of retail AR equipment. Original manufacturer parts are only available through them. The parts are not cheap, but then nor was the kit when it was new.

I'd also be wary of fitting a new part if I wasn't pretty sure I understood why the original part had failed.

Cheers,

GJ

[John Caswell]

Be prepared to dig deep for parts from Absolute.

John

[Radio Wrangler]

High voltage regulators are not circuits to be taken lightly, and extremely simplistic ones such as this are going to be, um, ticklish. I fear there is significant risk that this could keep repeating the fault.

As a commercial repair, I think I'd walk away.

David

Mea culpa on 1% LED series resistors

Fair enough, all we have is 1% too only stocked in E12 values though.