Audio Research Reference 2 Pre Amp Dead 1CH

[nigelr2000]

My mate with more money than sense has one of these pre-amps, it cost him £3000 second hand!!!!!!!!

He was always fiddling plugging different valves in to try for a "sweeter" sound, one day he unplugged a valve without first turning it off. There was a loud thump in the speakers eeeek.
He turned it off, replaced the valve, turned it on and was rewarded with one channel dead suprisingly not the channel he removed the valve from?

It then came to me.

I am at a loss how the electronic volume control operates. I get signal all the way up to pin 10 on U4 then i think it should emerge from pin 11 which it doesn't, I just get a voltage according to the volume control setting from 0 to 5 volts, comparing with the other channel it has only signal on it's pin 10 no voltage. I have replaced U4 & U5 seperately, U4 is a bit pricey so I wonder if the old U5 blew it up again ? I don't want to buy another one at £16 a pop only to blow it again so I wondered if anyone has any experience of using these digital potentiomers, are they as fragile as they seem and any othere ideas that might help me.

I have uploaded a pdf of the relevant page in the manual but hi-fi engine has the full manual.
Attached Thumbnails

[stacman]

Are the reels of sellotape around the valves for a reason.....
Regards

[Ted Kendall]

Anti-microphony rings...

[Radio Wrangler]

Think of them simply as a stepped pot.

Vhigh at the top
Vlow at the bottom
Vwiper from the selected tap.

Serial data fed into U/D, CS and INC select the tap.

There are TWO of these digipots stacked one on top of the other and U5 gets to choose whether to pass on the voltage from the top tap or the bottom tap to get a bit more range and resolution (It doubles the number of taps available so it really does increase resolution by one bit!)

More of those damned current source FETs And DC coupling in valve stages

But why bother with valves after using cheapie solid state analogue switches?

Sorry, but it's a hilarious design.

David

[dave cox]

Like David says, there are 2 stacked pots of 100K and 10K.
Ditto for the -ve input side of the balanced inputs.

I assume you are tracking an input signal to pin 10 / U3 ?
You should then see ~10% of this pin 9 / U3 and pin 10 / U4.
Pin 11 / U3 should be between 10% and 100% depending on the volume control (verify, A).
Pin 11 / U4 should be between 0% and 10% depending on the volume control (verify B).

If A is changing with the volume control but B is not, then U4 may be faulty.
If NEITHER is changing then the micro-control (driving 2/3/4) MAY be dead.

You should see some digital activity on pin 2/3/4 as you alter the volume control.

dc

[Ted Kendall]

It has been suggested quite seriously that the reported sound quality of hi fi gear is most directly related to its price. If I'd dropped three grand on something like this (as if!), I'd have hoped the manufacturer had got it right.

[dave cox]

From an engineering point of view, a single 5534 op-amp would put it in its place

[Radio Wrangler]

Originally Posted by Ted Kendall

It has been suggested quite seriously that the reported sound quality of hi fi gear is most directly related to its price. If I'd dropped three grand on something like this (as if!), I'd have hoped the manufacturer had got it right.

I think the mechanism is that anyone who has dropped three grand on something like this will be able to convince themselves that they can hear an improvement. If they'd dropped thirty grand, they'd be able to convince themselves even further.

But, no, if you look a the diagrams, there are a number of silly things done just to be different, just to be something the marketing people can point at. Do they make it better? No. Do they make it worse? Yes, in several respects, not least repairablity.

Designing electronics by fashion is a very poor second to using science and mathematics.

David

[GrimJosef]

Originally Posted by nigelr2000

My mate with more money than sense has one of these pre-amps, it cost him £3000 second hand!!!!!!!!

... I wondered if anyone has any experience of using these digital potentiomers, are they as fragile as they seem and any othere ideas that might help me ...

I've worked on one. It was the AR Anniversary Ref. Your mate got off lightly. The Anniversary was £24k list and not far short of £10k second-hand. If tyiou could find one.

It also used digital pots, stepped as David has said. I never did find out what had caused the original problem, although I suspect that some conductive debris may have dropped in through ther vent slots on the top of the pre and traniently shorted the valves' HT rail to the -15V rail which powered the chips (both rails ran on top of a double-sided pcb so qwere vulnerable to falling objects). It turned out that whatever the fault was had blown everything - all 8 vol control chips were, to a greater or lesser extent, toast. AR had carefully matched them to get very high-accuracy tracking of the controls. They had sets of replacement chips for sale but these were expensive. So I built a jig, bought some chips from eBay (at least one type was obsolete), and matched up a couple of sets myself. It was a faff, but it worked fine in the end. Whatever the original fault was it hasn't, as far as I'm aware, recurred.

EDIT: Dave has raised the possibility of the microcontroller being damaged. Your mate had better hope that it hasn't been. It would be worth scoping the control signals (make sure they're not being shorted by blown circuitry in the vol controls) to check they're OK before investing time and effort in the vol control chips.

Cheers,

GJ

[Radio Wrangler]

Just to check in another possibility, this is a volatile memory type with pseudo log resistor chain.

It is supposed to come on at the 10% tap before data gets loaded to it, but this could be rather a short period if the CPU is quick off the mark.

I checked because U've used the non-volatile Maxim digipots. They are EEPROM based and the EEVPROM part has a wear-out mechanism, they are only good for so many re-writes. EEPROM wear-out is why odometers are starting to fail on high mileage cars... they've been written to too many times and their lives used up. It's something to check when you see a digipot used as anything other than a preset.

I see it's just a 2-slope approx to logarithmic, like the cheapest crappy panel pots. For use in something that price, I'd want something better tempered.

"The well-tempered volume pot"? better for listening to not just Bach.

As said, see if volume changes have any effect along the signal flow path.
Follow the signal and find where it stops.
Check for activity on the digital programming lines.

David

[Craig Sawyers]

Hmm. I always wondered what the schematic of one of their Reference Series preamps looked like. Now I know.

Another possibility is that the hot valve-swapping could have seen off one of the unobtainium selected mosfets.

I'll grab the full manual.

Craig

[Craig Sawyers]

Come to think of it, it is insane that a product cannot cope with a valve being unplugged with the product powered. Valves age and their characteristics change, heaters go open circuit, and heater cathode shorts are not unknown.

Craig

[GrimJosef]

Originally Posted by Radio Wrangler

... I see it's just a 2-slope approx to logarithmic, like the cheapest crappy panel pots. For use in something that price, I'd want something better tempered ...

I don't think I'd be giving too many secrets away if I said that in the pre I worked on the controller doesn't step the pots one step at a time. The combination of two pots gives a much more finely divided control range than the hundred-odd steps that are available to the listener. Each of the listener's steps is made up of a pre-programmed number of pot steps and the result is much more smoothly curved than double linear.

EDIT: With pot selection the matching can be exemplary too which, given that there is one pot-pair for the signal+ side of each balanced channel and another one for the signal- side, actually does matter. That would be tough to achieve with analogue pots.

Cheers,

GJ

[Radio Wrangler]

Yes, of course the step at which the slope changes will be well-defined, so some correction will be possible, but quantised to the step size, of course.

I'm just a little disappointed that Maxim didn't scale their resistors properly for a true log set of steps.

My preamp in the lounge uses a silver stud fader with individual 1% metal film resistors in feedback around an NE5534, with resistor values calculated to give equal steps. This seems to give the best of most worlds. A well-defined law, low noise resistors and the benefit of the active gain control in handling noise floor and compression ceiling.

Anywy in the preamp schematic for the unit under discussion, the analogue switches and the digipots do all the work, the valves are just for set-dressing. Has the signal been through a valve? Yep! Tick!

David

[dave cox]

This amplifier input stage looks like something I knocked-up in a full differential amplifier some years ago.

A pair of 'long tailed pairs', with common-ed emitter resistors, where feedback is applied to one LTP and the input applied to the other (its a standard mixer arrangement with a tiny twist). I used medium voltage BJT's throughout so it cost a little less than $10k, more like $10 It seemed to be quite difficult to get stable but I did see ~100V p-p before it got really ugly.

Did they miss a neat way of altering one of the tail currents (or an additional LTP) to get an electronic volume control without having to match 4x switch networks and without the signal going through all those ladder networks and FET switches ?

dc

[Radio Wrangler]

A transconductance multiplier, like the Gilbert cell? It's difficult getting those very linear. There is an improved variable gain amplifier based on it, the X-Amp from Analog devices (Barrie Gilbert works for them)

But the ADI part is several dollars, maybe ten times the cost of the Maxim part. The cost of the preamp would have to go up. They certainly wouldn't be able to keep it under $10k.

But seriously, a switched potentiometer network or ladder network would be the way to go. Switches can be made a good bit more linear than variable gain amplifiers.

Out of the box alternatives would be converting to RF, using PIN diode attenuators and then converting back to baseband, maybe that's in the domain of $100k preamps? It's always nice to be ahead of the game with the next trick up your sleeve.

David

[nigelr2000]

Thanks for all the comments I think I have a better idea of what they do and can test further. The digital bits seem to be working ok as I get bursts of data on the relevant pins when the volume control is operated.
I am hoping to get it back on the bench wednesday, I have a jukebox to finish off tomorrow so will try more tests and get back to you. I assume that maybe the output side of one of the chips may have a short and that why I get a variable voltage on the output.

[Craig Sawyers]

Originally Posted by Radio Wrangler

A transconductance multiplier, like the Gilbert cell? It's difficult getting those very linear. There is an improved variable gain amplifier based on it, the X-Amp from Analog devices (Barrie Gilbert works for them)

David

Did. Alas Barrie Gilbert passed away on 20th January this year aged 83.

Craig

[Radio Wrangler]

I must have missed that piece of news. Very sad. So good an inventive designer that ADI opened a division near him. It's quite a number of years since I last saw James Bryant as well.

David

[GrimJosef]

Originally Posted by Radio Wrangler

... But seriously, a switched potentiometer network or ladder network would be the way to go ...

For £24k today's customer expects to get a remote control. For everything. They're also sniffy about being able to set the volume at just the right level. So a 24-way or even a 36-way switch wouldn't cut the mustard. They want auto soft-start too, and with all the settings which were selected via the remote remembered for the next switch-on.

I suppose it's no surprise that modern designers seem to assume a microprocessor-based control system will be the default, and once that decision's been made chip-based volume control becomes the easy off-the-shelf option. I'm told the chips have come on quite a long way since the noughties, when the pre I worked on was designed, and these days there would be no need to pair them up to get the required number of steps and 'taper'.

Cheers,

GJ