Revox B252 Preamp
 

Well, I've finally got one. It was bought in non-working condition from an auction site.

The unit arrived beautifully packed and is cosmetically excellent. The fault manifests itself exactly as described. My brother has all the right accounts set up etc and helped. We had a bid in at what I'd really be prepared to pay up to and it did the deed. Of course, it was coming in well under that until someone sniped in a bit that lifted it a hundred euros with 15 seconds to go, but the automatic bid trumped him by a small increment and there was no time for him to have another shot. So I got it within what I was prepared to pay and my first ever experience of an on-line auction went as well as I could have hoped. The seller got a good price for it, considering what working ones fetch.

Anyway, this is about fixing an interesting bit of gear, though the purchasing was an entirely new experience for me. So now I have to put my skills where my mouth is.

Job-one was to twiddle the mains voltage selector switch to 240v and try it with a bit of power.

The B251 integrated amplifier and the B252 preamplifier are identical looking boxes. The B251 has a very crude switch mode power supply and a pair of power amps in addition to the preamp boards. To make the B252 Mr Studer's merry men whipped out the power amps and their weird heat-pipes and heatsinks along with the SMPS, and stuffed in a nice iron mains transformer and simple linear regulators.

The same manual covers both models.

I have a nice power amp of my own doing, and only wanted the preamp. I'b been prepared to get a B251 and convert it to a preamp by making a new linear PSU.

I've had a general look around inside. There is a nice coating of undisturbed dust so it looks free of the trails left by the phantom. The reservoir capacitors look a bit old and tired. One seems to have been replaced long ago.

Applying mains illuminates the subsonic filter switch LED, which does douple duty as a standby indicator. This is how it should be. But the tone control defeat button LED lights, some bars (no words) appear on the LCD and the backlight lights very dimly. Too dimly.

This unit uses a 12v festoon bulb, while some other models in the series use 35v. I wonder if the wrong bulb has been fitted, or is the 12v supply running very low?

Pressing the power on/off button has no effect. No controls have any effect. I don't have the IR remote control to try that.

That's all for now. I'll have some time on it next week. If anyone's interested, I'll post photos and document the debugging trail. I can also do some photos of the sorting of the matching B261 tuner.

I suppose someone could try the usual auction sites and find an item which may have been this, but I'm 100% satisfied with my purchase. Mr Sniper was anonymous, though I did get to see his recent bidding history and it was all last minute stuff, single bids, mostly unsuccessful and spread across all sorts of things. It's probably how he gets his kicks. In the end, the price was fair to both vendor and purchaser at about 2/3 that of a working example.

David

Re: Revox B252 Preamp
 

Bidding on ebay can be tense and exciting, it was better before the auto bid, I've got a few bit's by waiting till the 10s countdown. Happy to hear you finally got one David.

Am interested in the repair, be nice to see how you go about fixing it.

Andy.

Re: Revox B252 Preamp
 

No further discussion of eBay please.

Re: Revox B252 Preamp
 

It's a bit of a beast! Rather more interesting that the usual hi-fi fare. In terms of complexity, it's like a professional product (funnily enough ;)). Very discrete - hardly any op-amps. I wonder why - the NE5532/NE5534 were well-established by that time (indeed, there are some in there). I see the input buffers are fed from +/-25V, though I'm not sure what source devices could make signals big enough to require that sort of headroom...

If they'd used some other form of input switching, they could have done away with the input buffers, but as they used JFETs feeding into virtual earth buses, the buffers allow them to keep the resistance values nice and low to minimise noise. I'm working on a similar thing at the moment, but I'm doing the opposite: series-switching into very high Z (to minimise distortion) but fed from very low Z to minimise crosstalk. Using solid-state switching is always complicated, and they usually set the noise or distortion performance in an otherwise "blameless" design. There's a lot to be said for relays ;)

Surprisingly, the AD7528 M-DAC is still available. Respectable performance, but today a PGA2310 is a better way to control volume. Nothing like that existed in the early '80s, of course.

Sounds like a "simple" logic or PSU problem. Is the CPU mask-programmed, or is there EPROM in there?

Looking forward to seeing the progress reports :thumbsup:

Re: Revox B252 Preamp
 

There's many a true word spoken in jest - I've run up against the same problems. That said, Quad made it work in the 34/44, largely by restricting the available voltage swing. My Focusrite router does it with relays, very well, and handles large input levels with aplomb. I have read and re-read Self on the subject and ended up wanting to scream...

Re: Revox B252 Preamp
 

Yes, I'm always amused when folk say there's gain ahead of the switching in those Quad preamps because they haven't spotted that the op-amps are being used as voltage limiters, and their gain is actually unity (or less for some inputs that attenuate ahead of the op-amp). Some versions of the 405 do the voltage limiting that way too. It's a technique I'm using in a design I'm working on now with 4052 CMOS stuff. Funnily enough, I don't think Self covers it in his discussion of voltage limiting. The only downside is the poorly-defined saturation points of the op-amps.

I have "Small Signal Audio Design" on the bench right now. He dismisses voltage-mode CMOS switch rather too easily, clearly preferring current-mode (as used in this Revox). But if you're doing a router, current-mode becomes noisy, or you need seriously Herculean input buffers. I'm doing a 8 by 4 one here (small by broadcast standards), using the DG407. Results so far are promising, but not as clean as relays.

Re: Revox B252 Preamp
 

Yes, it does seem to have massive headroom.

My current preamp is a rat's nest built thing I did in 1981, built around NE5534s. The mm cartridge input was an NE5534 circuit by Peter Baxendall, the one with an added passive pole on the tail end. Switching was voltage mode JFET with bootstrapped gates, a circuit suggested by Bill Miller before he moved to Linn. The switches work well, but reliability of the very high gate bias resistors (22 meg) is its Achilles' heel. I keep having to replace some and I'm getting fed up. Hence the decision to get a B252 to match my B261 tuner and A77. If in the end I don't get on with the fixed B252, or some unobtainable part is needed, I'll just design my own replacement boards for it and keep the case.

It has two mask programmed CPUs, no EPROMs One handles the remote, the other handles the LCD PPMs and I'm not sure which handles switching... the LCD one looks likely. I wonder if the masked EPROMS are the same as in the B251 with the power amp bits parked?

With the backlight so dim, I'm hoping for an easy route to a PSU fault and a low rail voltage, but it could be that someone put the wrong bulb in (other Revox products use higher voltage festoon bulbs.) and the dimness is a red herring.

My tuner was rack mounted, so the plastic side panels are missing. I made solid oak ones. So I either need to make another oak pair to make the B252 match, or go looking for two plastic sides for the B261.

My power amp needs 0dBm/600 Ohms for full welly,, so I need to attenuate the preamp outputs and I'll want to steal switched 12v for the remote mains switch relay in the power amp. Noise-wise it would be smarter to attenuate at the power amp end.

The plan for Monday evening is to start by checking the supplies from the PSU board, and check the crystal clock oscillators are going.

David

Re: Revox B252 Preamp
 

OK, home again and a first poke inside.

The festoon bulb does indeed have 12 across it, so the dimness is from the wrong bulb being fitted... but no, the bulb on one endcap says 12v 42mA. Odd? so why so dim? I can only just tell the LCD is lit. Anyway, I dropped in a spare LED festoon from sorting the B261 and it's an awful lot brighter. I'll need to go to a lower power one. It helps that these things beam in one direction unlike the filament bulb.

All the power supply rails are up and running.

The transformer, rectifiers, reservoirs and 5v regulator run continuously and the amplifier has standby and on modes only. %v runs the CPU and IR receiver waiting for the 'go' command. Transistor switches gate off the rectified 35v from the +/-25v and +/-16v regulators. The control line says on is low. I measured +20mV on the line, and the PSU has turned on as it is supposed to. The light is powered from the unreg side of the 5v regulator and is gated by a bipolar devive turned on if the +16v rail is alive. This makes the LCD lamp go out in standby.

The heatsink which does all the linear regs runs quite warm.

The on/off switch button makes no difference to the PSU enable line, it stays low (=ON) and pressing the button doesn't change it.

Enough for tonight!

David

Re: Revox B252 Preamp
 

Another poke inside:

Well, it doesn't look like an easy PSU fault. I've noticed that on turn on, I get a momentary flash of all the bars and annunciators on the LCD, then it goes blank, then about a dozen bars appear for each channel, with nothing else showing.

I can get to and probe the main CPU easily. There is a second one called 'remote' which handles the IR receiver and scans the keyboard. The main CPU runs the ADC arrangement for the Peak Programme Metering and communicates with the input selection switches, volume controls etc.

The crystal connected to the mainCPU is oscillating with a clean sine at about 4MHz, which looks right. There are data and clock output pins which are directed by a demux switch chip to go to various places, There is activity, 0-5v pulses on both of these, in periodic bursts, though the address lines to the Demux are static. The reset pin is sitting at about 1v. Interrupting the mains and re-connecting sends it up to 5v and back with a fairly slow dv/dt this doesn't seem bad.

There are pins for a DIY adc which are staying high as are pins to what looks like an EPROM. There are ready and ack lines to/from the second CPU, obviously part of some handshake. Ack staus high, but ready is low with some short pulses to less than +1v. Either these are glitches or something can't drive +5v.

So the main CPU is doing things. I can't get to the remote CPU without more dismantling.

I get the feeling the main CPU is waiting for something.

These are mask programmed CPUs, so what is the EPROM for, or is it a last-state memory so the thing comes up as you left it? It would be inconvenient if it didn't.

There is a little add-on board near the crystal and reset pins that carries one IC. It's not in the manual from Electrotanya and it's mounted on end by some wires into the processor board. It's wobbly. I wonder about broken tracks. I wonder what it does?

That heatsink definitely runs warmer than I like. Is something taking too much current? There's no feel of lots of heat from anywhere.

I think getting in to look at the second CPU comes next. I wonder whatit does for a clock? there's no continuous 4Hz coming from the main CPU.

The clock/data distribution switch feeds what looks like a shift reg as one destination. One output of this is a low to turn on the PSU. Now this shift reg has its own reset circuit. The PSU on line is low for true, so reset happens on mains application (the 5v rail runs continuously even if the preamp is turned off) so after a mains interruption, the power supplies will come on until the CPU wakes up and determines whether they should beon or not. So it looks like the power on all the time mains is applied isn't a fault, but is a further symptom of the CPUs not wanting to play.

I'm just thinking aloud.

David

David

Re: Revox B252 Preamp
 

The "remote" CPU processes local key presses... Looks like it's own local Xtal.

Doesn't page 74 of the PDF (right hand side) show the "reset PCB"? Admittedly that's the "amplifier" rather than "pre-amplifier", as shown on page 121, but perhaps that's still applicable?

Plenty of shift registers there expanding the output capability of the main uP. The demux 4052 that is being used to direct clock and data to multiple destinations, including the keyboard/remote CPU is to be viewed with suspicion - I've had more than my fair share of problems with those over the years.

It's not the easiest of manuals to follow on the screen - might need to get some A3 print-outs done tomorrow. I've spent a fair bit of time staring at the various photos of the innards that I've found online, and it's slowly starting to click. Would be easier if I had it in front of me. But getting there...

I've had enough scrolling up and down for one evening ;)

BTW, have you got a DSO? Guessing you probably have. Incredibly useful for this sort of thing...

Re: Revox B252 Preamp
 

Thanks, Mark

The reset board in the manual looks quite different to the one in this unit. The one in the manual seems mounted horizontally over the crystal with long wires extending down to the main CPU board. Mine has a single IC on it, is mounted edge on to the main PCB with short wires and has nothing to support it against hinge-mode flexing. I need to get the main board out and look for damaged tracks/pulled pads. It feels too loose.

I'm managing to follow it on-screen OK, but it's tedious squiffing around with the magnification up high.

Tried downloading a manual from the Revox part of the studer site. Others tried worked, but the B251/B252 fails to download.

I've always avoided designing with 4000 CMOS. I suppose some of these must be from the purple plague years. Dato on the 14084B presumably motorola MC prefix to go with that leading "1" seems hard to find and I hate those clickbait datasheet sites... could I have mis-read, the * isn't very clear.. MC14058B maybe?

Yes I have a DSO but haven't needed to break it out yet, the old 465 is finding activity or lack thereof well enough so far. I've even got an actual Philips logic analyser which might come in useful I suppose, but interchip comms is serial and RAM/ROM is internal.

I'll just spend the odd hour plodding along. If worst comes to worst, I suppose I make a new board and have to write software from scratch.

David

Re: Revox B252 Preamp
 

I thought you'd have it fixed in an hour or so, being an amp and not a HP spectrum analyser. It's nice to see boffin's struggle too fixing thing's. (Don't mean that in a nasty way)

Following some of what you say and taking mental notes. Keep up the good work.

Andy.

Re: Revox B252 Preamp
 

Hi Andy,

This device has two microprocessors controlling it. The software they are running isn't soft at all, it was built into the upper metal layers of the chips when they were made. The semiconductor manufacturers charged a huge tooling-up fee for each batch and wanted a huge batch size as well. And this preamp has two of them! (shared with the B251 integrated amplifier version). It's likely every processor Revox bought was made in one batch and eventual exhaustion of their stock was what forced a change to a new model.

To say these parts are unavailable is a bit of an understatement.

Surrounding the pair are several logic ICs, just gates and flip-flops inside. These are from an early family of CMOS - complementary MOSFET logic which was originated by RCA. At the time it was known to be fiercely sensitive to static charge caused failure. Later we learned that a glass coating over the chip would leach out phosphoric acid etch away the chip's aluminium connection layer if moisture got in. Moisture got in. This stuff was so efficient, many chips not running fast lacked the heat to drive out moisture from the atmosphere and years later they died.

I want this machine working to go in place of my old homebrew preamp in the lounge. It will match the B261 tuner and A77. These things sell for a few hundred pounds not working. Workers fetch about £700, and ones where someone replaced all the capaci... er 'restored' it can be found with asking prices around £1300.

I bought it with a declared fault which made it certain that the processor system was down.

The fault was exactly as the seller's description, and I understood what it meant. No-one seems to have ''had a go' at it before, it's all original with no marks in the film of dust.

My first hope was that a simple fault like a dead power supply rail was stopping it... the dim back light fitted this. But all supplies are running. There is 12v at the bulb as well and the bulb says 12v 42mA on it. It seems very dim though. Spike Milligan joked of living in digs where you turned the light on and the room got darker.

The next hope for an easy fix lies in finding a bad solder joint or a broken track around that add-on board.

Then comes the possibility of some dead 4000-series CMOS logic. Some of these cam be hard to find and the purple plague corrosion kills NOS ones sitting in stock. A bit of deft substitution with a 74HCxxx series part is possible, but pinouts are different. I could put an SMT part on a snippet of PCB as an adaptor.

The nuclear option is if either CPU is dead. All bets are off. I'll need to make two new boards for the processor system based on different processors. I could use an arduino or raspberry pi with a layer of interface adaptors, but I'd more likely use a medium size FPGA. Some have internal CPUs and the soft logic can be programmed to replace all the CMOS parts. But this is a fairly simple system, I could write my own processor out of FPGA logic cells and write my own code for it. I've done something like this before and it can be fun. This offers the possibility of a bit of commercialisation... get a small batch of boards made, load them up, test them, keep a spare set for myself and flog the rest to offset costs.

I'm not in a rush.

One of the logic boards is not accessible with the thing together. Whatever your status on the boffin scale is, fiddly is still fiddly. Studer's boards and connectors of this period are fragile. And quick diagnosis still involves luck. I'm not stuck, but I am getting ready to engage low gear. Slow but unstoppable. (The analogy wasn't lost on me on sunday when a yodel driver had reversed his van off the drive in the borders and spun his wheels til the chassis grounded. Yodel man number two in his van couldn't shift it with a tow rope. Low ratio in my beastie did it easily and I still had the difflocks in reserve)

David

Re: Revox B252 Preamp
 

It's a 4094 - an 8 bit shift register with output latches. These are liberally sprinkled throughout, and are a dead-simple way of getting lots and lots of outputs from a microcontroller, using just 3 of its pins. Today, the 74HC595 seems to be the preferred option, but the 4094 is still in production and readily available (e.g. Farnell 1106118).


The advantage of a DSO is capturing one-off events. They are especially good for serial stuff. Many will even decode the serial into ASCII for you (probably not all that useful here). In a situation like this, you're looking to see what happens at power-up, to establish what parts are being spoken to. For example, hanging the DSO across the inputs to IC4 on the Processor PCB would be really interesting - is the shift register ignoring input data, or is it not being spoken to? You're expecting to see 8* clock pulses on pin 3, some changing data on pin 2, and a pulse on 1 at the end to transfer the internal contents of the shift register to the output latches.

* Of course, shift registers can be cascaded (see pin 10), so you'll have to wait for several multiples of 8 before pin 1 is pulsed. Here, pin 10 of IC4 is going to the Volume PCB, which leads to another 4094 (IC701) and another 2 on the Supply/Output PCB (via a ribbon plugged into the socket of IC702). So you should see a total of 32 clock cycles before the latch is asserted.

Of course, if one of the shift registers in a chain was to die, then subsequent registers won't be getting their data...

With a modern DSO with reasonable memory depth, you can predict what each output pin across all shift registers will be just by looking at the state of pin 2 with each clock pulse.

This idea of daisy-chaining a whole load of shift registers is very common. Great for things that don't need to change very fast (a microcontroller port can be clocked at MHz speed if desired - often not needed). Also great for saving wire and PCB traces.

If you're feeling really cheap, you can do away with output latches - just 2 wires to route. Think about those awful Tek 2465 things - have you noticed that when you press a button, all the LEDs on the front panel blink briefly? That's because they're driven via shift registers that don't include the output latches (74F164), and presumably don't clock them all that quickly. So what you're seeing is the updated patten of LEDs "rippling" through. The "fanboys" don't seem to notice, but I saw it straight away and thought "d'oh!" ;)

Anyway, the downside is that every time stuff needs updating, a large "burst" of data is made, which travels everywhere in the instrument, potentially causing low-level audio interference. So that's why they are using the 4052 mux - those noisy clock and data signals are only going around the audio boards when the system needs to change their settings. Indeed, they've separated out volume and input selection. But because we have 2 microcontrollers chit-chatting - potentially quite frequently - the 4052 ensures that inter-processor chat does not result in the clock/data signals flowing around the audio PCBs. Neat - nice attention to detail - but something else to go wrong.

Of course, if they'd chosen a micro-controller with more output pins, they could have avoided the 4052 mux altogether.


I'd use a PIC. Rather like this: http://www.markhennessy.co.uk/preamp/

If you look at the schematics you'll see plenty of commonality. I used serial control for input selection (latched shift registers with integral Darlington drivers), volume control (PGA2310s, which have latched shift registers inside) and 12V triggers (a 75HC595 and driver), plus a 74HC595 for the LEDs on the input selection PCB. I didn't do as much cascading as I could have done, but there's some. The code for that is pretty trivial, BTW. Obviously, my preamp has a much simpler audio path ;)

A pi would be overkill, unless you want to add AoIP functionality ;D

Hope this helps - apologies for where I stated the obvious, etc...

Mark

Re: Revox B252 Preamp
 

David

I'd go for a PIC too, there's plenty of choice of devices with all sorts of I/O options and variants that run from a 5V supply which is less and less common these days. They're dirt cheap too.
I strongly dislike PIC assembly language, but high level language compilers are available. C would be a good choice if you're familiar with it.

John

Re: Revox B252 Preamp
 

You have us all looking at the manual now! At least the processor is running which raises the probability of the fault being outside. If you are looking at the same service manual there is no reference to any self-diagnostics or how to set the EEPROM back to defaults.

The obvious cause of a lockup would be communications to the other microprocessor as this would involve a communications sequence. Most other devices such as the EEPROM might return the wrong data but not cause it to freeze. I would guess the interface micro is most likely to be the master.

Re: Revox B252 Preamp
 

I've not got down to the 'remote' CPU yet. Tuesday evenings are usually spent on horseback, but I wouldn't have got home early anyway with another impromptu closure of the Forth Road Bridge. I'll try to get at it tonight.

The data line in the link to the 'remote' CPU (which means the one which also interfaces the infra-red remote) was seen to be sitting low with sub-1v runt pulses on it. I want to find out if the remote CPU is running or whether something is loading this line. I also want to get a look at the underside of the main CPU board where I suspect there is a damaged track.

As far as which is master, these processors have room for only a scoddy amount of code, so I suspect tasks got shared out to fit. 'Remote' got the IR input, keyboard and LEDs, 'Main' got the metering, volume control and the little Eeprom. The Eeprom has a lifetime of about 100,000 write cycles. I guess it holds state when power is off, and not a lot else.

I don't want to get involved in alternatives unless something unobtainable is kaput, and then it would be a matter of choosing something I'm comfortable with. VHDL rather than C. I first learned Fortran II using card stacks submitted to acolytes with an ICL mainframe! Everything was easy and convenient after that.

The Revox manual just gives schematics, material lists and no testing or modus operandi. Not one of the most helpful. Though it does show that they mucked around with changes to reset circuits... hmmmm?


David

Re: Revox B252 Preamp
 

As far as I can see, the only unobtainable parts are the mask-programmed micros and the EEPROM (and possibly the AD7110 used for the meter ADC). Everything else should be available new, and NOS might be an option for the latter two parts. Mask-programmed microcontrollers seem to be extremely reliable. Who was it that said "big chips don't fail" in Television Magazine? No need to panic yet...

If the EEPROM wears out, the code should account for that and assume sensible "defaults" at power-up. "Should"...

I can't find a datasheet for the MAB8440 yet, so don't know how much code memory they had, but you don't need much for this sort of thing. After all, there's no text strings or bitmaps stored in there. A few hundred words at most... My PIC preamp started out with a 1k chip, and that only ran out of memory because of the menu system..

So the need for 2 CPUs might not have been memory as such - rather, it might have been about resources. Decoding IR remote control is simple, but requires relatively precise timings - it's possible that trying to deal with the level meter at the same time was asking too much - looking at how they've implemented the ADC, it will require precise timings as well. Ultimately, only the original design team know the truth, but it's interesting to wonder...

Personally, I love PIC assembler. I've never liked C, and especially not for simple embedded applications like this. My experience of programmable logic is minimal, and is nowhere near VHDL. If programming up an alternative device, the hard decision will be how near to a 1:1 replacement would you make it? All that logic could be massively simplified - probably down to a single PIC16F877 or similar - but then it's not a drop-in replacement that can be sold to others with this fault (assuming it's a common one). But we're getting well ahead of ourselves (because it's fun and interesting!) - it probably won't come to that...

I suspect that once you've got it apart a bit more and continued with the common-sense checking, it'll yield pretty easily.

Re: Revox B252 Preamp
 

Here's an outline MAB8440 datasheet

http://datasheet.datasheetarchive.co...DSAP005622.pdf

these devices appear to have 4K bytes of ROM and 128 bytes of RAM.

John

Re: Revox B252 Preamp
 

Thanks, guys,

I've designed some fairly big Altera FPGAs for doing serious signal processing, so that's what I'm comfortable with... I also put an array of several fractional-N synthesisers onto one (US pat 6509800) to try an idea for a bit of fun.

I could do the thing as a pure hardware machine, but I'd still want something non-volatile to hold last-used settings.

Anyway, as Mark says, I haven't got to the bottom of what's wrong, so I considered worst-case possibilities before I bought it and reckoned it was still worthwhile.

Being mask programmed, only an electron microscope would get the code out, and I don't have a working one for comparison, so any writing is from scratch. If I have to change cpu type, then I have to do a new board, so the PPM ADCs get done from scratch. No big deal.

CD4000 stuff has a fatality mode, the bigger chips are probably a lot better and it may be mechanical damage from vibration etc.

Yup, this is fun.

David