Well, it may be an expensive 'experiment'....only time will tell I suppose, but the guy that did the work does seem to know what he's talking about (and he does have a machine himself which he primarily uses to x-fer stuff from 1" to digital for clients). Of one of the cards (CH10) he did say this (you may be better reading from the bottom of this posting as I've copied his remarks with the most recent first):
Recording tests passed! Just waiting for the final few capacitors (promised tomorrow) to get the replay bass response to spec and they'll be ready to come back.
I've had the cards on soak-test for 3 hours today, so I'm as confident as I can be that I've cleared all the intermittent issues. I've calibrated the meters and input record level at TP3 as per the service manual, but beyond that it's over to you to calibrate to your head and bias to the tape you choose to use.
One issue that I can't fix: I'm suspicious about the Dolby circuit on channel 10. With NR 'off' the channel plays as expected, but there are some unexpected dips in the frequency response with it 'on'. There are two sealed calibration pots for the Dolby circuit which require specialised test gear to set up,
but I suspect it's the module that will need replacing - not a part I have access to. You'll need to contact Fostex for advice, or possibly find a spare card somewhere.
As you may have gathered it has been an absolute nightmare to trace all the intermittent issues - I'll enclose a full service report with the boards. I can only think I must have been incredibly lucky with the machine I bought - no further faults at all after the electrolytic caps were replaced.
I suggested swapping CH10 over to CH24 as I will be using timecode which doesn't need Dolby
Yes, swapping cards 10 and 24 would seem to be a perfect solution.
The problems are 99% down to Fostex using substandard electrolytic capacitors - it's a known problem with these machines. Think of cheap unbranded batteries that leak as opposed to Duracells that don't. Now they've been replaced you should have no similar issues, not for many, many years anyway. You'll see that the replacements are three times the size of the originals: these are quality components, chosen for maximum reliability and performance. The boards don't look so neat, but you'll get the best possible life out of them. The circuit is not power-hungry and the cards don't generate much heat, so no cooling required.…
Hmmm… Sorry about that - very annoying - I calibrated the meter drives and they were all working fine here. Odd that it isn't reading full scale: if the op-amp chip fails you'd normally get full output voltage or zero volts. The meter drive design is a little unorthodox to say the least and very sensitive to leakage current, so perhaps there's some electrolyte left that I didn't manage to clean off.
One little test, if possible please. Can you input an audio signal that is enough to just read full scale when input monitoring on a working channel, then apply this same signal to the rogue card input and see if more LEDs light. If so, then leakage is to blame, and a gentle clean with isopropyl alcohol and a soft toothbrush around U5, D8, R68 and R69 and C42 - on the square inch or so just above the larger edge connector - may do the trick. Dry off the board with cotton buds and rinse and dry a second time with a drop more alcohol. If not, it could indeed be that it's another chip gone south or another diode breakdown.…
I'd agree with you there - something odd on the card. However, it's the bit about the meter reading signal only when the output lead is plugged in that makes absolutely no sense - it suggests that signal is being introduced along the cable somehow. For example if you take the input lead carrying the 1kHz tone and accidently plug into the output you will get a meter reading if the signal level is high enough…
I can't think of anything else so I suggest replacing the meter drive op-amp and cleaning the contacts underneath is probably the best way forward. By all means post the card back to me and I'll look at it as soon as it arrives. Sods law will, of course, mean it will work perfecly in my machine, but I'm happy to replace a few parts on the off-chance if the fault doesn't show itself. The joy of intermittent faults!
Your board is on the way back.
True to form, it initially worked, but fairly quickly the meter started reading without any input. It turns out the output buffer op-amp was unstable and oscillating at around 270 kHz. This is a problem with most op-amps and highly capacitive loads, hence the output cable triggering it, but there are components in the circuit to maintain stability which should cope with anything other than impractically long cables. I suspect good old electrolyte leakage was to blame. I've lifted and cleaned all the associated components and replaced the op-amp IC, and normal service has been resumed.…
Oh, ***k! Very sorry you're having further issues. I'm beginning to wonder if these output op-amps have all been stressed somehow by something like a power surge in the past - there's no reason why they should fail now. The lack of input/output could also indicate failure of the 5V regulator, and these certainly could have been stressed by the leaky capacitors, or a power surge. I replaced 3 if I recall. Can you double-check the +24V, +15V and +12V power rails, please.
Yes, I think some serious soak-testing is called for - the 3 or 4 hours I had the machine running for would seem to be insufficient, though I repeat, there's no good reason for such components to fail now the leaky caps are gone. Having replaced all the electrolytics in my machine (and nothing else!) it's worked flawlessly ever since.…
'There is a 'logic' connection to each audio card, so it's possible that a bit of bad data in memory was causing switching issues, particularly if the memory battery is either failing or insufficiently charged. If audio is at the output but the meter isn't reading though it does point to the op-amp IC. Please take a look at card 21 and see if the 8-pin chip next to the group of 4-pins of the Dolby sub module has been replaced. Also check the 5V rail on this card if possible - yes, remove the 4 of 5 boards next to it to gain access.
Well, at last there is daylight appearing at the end of the tunnel that has been this week! It's been frustrating as I've had to clear faults in sequence, which has meant ordering components in different batches as required.
Having cleared the failed op-amps, most of the issues I've encountered have been due to contamination of high-impedance stages of the circuit. For example, the switching FETs have floating gates when 'On' which have proved to be incredibly sensitive - breathe on the board and they switch 'Off'! It seems the combination of inadequate cleaning of flux at the factory, coupled with electrolyte leakage and possibly a hint of nicotine has been a very bad combination. Fortunately a years supply of cotton buds and swabbing alcohol has got them switching reliably.
I've had to replace two 5V regulators that had failed, but fortunately no further damage to the logic circuits they power had been sustained.
The playback amp also has a high-impedance path, and I'm currently waiting for some more capacitors (responsible for the correct replay EQ) which have absorbed electrolyte, causing loss of bass frequencies and increased distortion. This, with everything firmly crossed, SHOULD be the final step!
These caps should be here on Tuesday, and I'll be able to fit and final test then. There is evidence that pots have been twiddled with, so I'll calibrate the meters and replay EQ, which will be a good starting point for final tweaks to match your replay head.
An op-amp (operational amplifier) is an integrated circuit (a silicon chip) - like a transistor but with 8 legs, in this case. They're more awkward to deal with than capacitors because of the more complicated soldering. Semiconductors are the active elements in circuits and hence the most prone to failure - resistors almost never fail in low-power circuits like these.
So far I've found about 10 of the same device in the same circuit location which have failed on one half or the other (each 'chip' contains two amplifiers). They're not expensive - around £1 each - so it's crazy not to replace as needed. These failures account for such issues as no audio output, dead meter or full scale meter reading, depending on how the device has failed. I need to replace these first before I can check the earlier stages are working correctly.
All in a day's work!
(In response to finding a fair few op-amps failed he asked me to test three voltages on my machine, which reported in spec)
'Good. It's worthwhile proceeding with the opamp replacements then.
I'm not worried about the meter readings with the cards removed; the way the meters are driven in a bit of a fudge to be honest - hardly a precision circuit design, and there seems to be a fair bit of 'OK, that'll do' attitude evident!
With so may semiconductor failures powering is the first thing to suspect, hence the need for the test. It could be that a previous owner was removing/replacing boards with the machine powered up - guaranteed to kill something.
I'll plod on…
Hi Alan,
I've found a fair number of blown op-amp chips so far, mostly the line output/meter drive ones, but there are some other odd faults I've yet to trace. It seems like a lot to go wrong, and far more that I'd expect from leakage issues. I'm beginning to suspect that the machine has had a power surge at some point. Can you access a decent quality multimeter and verify that the power supply +24V, +15V and +12V rails that feed the audio cards are all within a fraction of a volt, please? I'd expect the tolerance to be + or - 0.2V on all rails.
I've attached a PDF of the power supply circuit and connector board: the voltages are on connectors J2 and J3 on the regulator board, destined for the main connector board which the audio cards plug into, or, possibly easier, on J7 4 on the connector board itself.
I had to re-order the chips as they sent the wrong package - too small even for these boards, would you believe!
Sorry about the slow progress.
Just a quick update. I've completed the cap replacement and have the boards installed in my machine for testing. Unfortunately five of the boards have op-amps that have failed, most likely because of the leakage currents. I have ordered replacements and will fit as soon as they arrive.
It does seem rather strange that the affected boards are 4, 8, 12, and 16 - a bit of a sequence - plus one odd board, 17. Looking at the circuit diagram it seems unlikely that a failure elsewhere could have caused these chips to get fried, so to speak, but it is an odd coincidence. Can you please power up your machine briefly without the audio boards, and see if there are any residual meter readings.
OK, that's somewhat encouraging - I've seen much worse! You're right - the erroneous meter readings show that the capacitors are leaking badly (electrical current, as opposed to physical electrolyte, though that's the next stage). The only way to get a reliable machine that performs to spec is to replace the lot, as they will all be of a similar age and state of decay and will only get worse. We're talking about the rectangular plastic cased electrolytics with both terminals at one end(e.g. C21, C22, C27, C28 etc.), not the ceramic 'axial' types or conventional through-hole types.
In theory, assuming the preset pots haven't been twiddled in an attempt to equalise the meter readings, the boards should just slot back into place after cap replacement and you'll be in business: there should be no need to ship the whole machine.
Assuming all boards are as clean as the photos and you send me just the boards, to replace all surface mount electrolytics with top-quality Panasonic audio ECA, long-life low impedance EE series and polyester film as appropriate, time taken between 1.5 and 2 hours per board, plus components at cost:
One caveat with these machines, though. It's most likely that you'll be using a different type of tape to that which it was calibrated for in the factory (Ampex 456, to the best of my knowledge, now no longer made, and most of the old-stock will be 'sticky' - avoid!) This re-biasing is an absolute pain on single-head machines, since you can only do it incrementally and not in real time: record a tone, play it back, adjust a bit, record another tone, play it back, adjust a bit more, and so on. Repeat for each channel… I assume you have the necessary test gear to do this.
You'd be wise to purchase a reference tape to correctly calibrate the playback response before attempting to adjust the record side. Canford should still have them: 320nWb, 15ips, IEC-1 is the type you need.
Out of curiosity: are you intending to record new material on this machine or is it purely for transferring archive tapes? Frankly, there are more economic stand-alone digital recorders available - the Alesis HD 24 being my favourite tape-machine replacement.
I do all work in-house. This was Fostex's first foray into surface mount and their chose of the miniature electrolytics was a very bad one - they all end up leaking (both current and electrolyte.) Best solution is to replace all (30+ per channel) with full size non-surface mount - I use special audio-grade caps. There is enough room between the boards to accommodate these components, you just have to be a little more careful when sliding them in.
The biggest problem is where electrolyte has leaked onto the board - the tracks are very thin and I had to get into some serious cleaning before the new components could be fitted, even repairing a few tracks.
Please extract a couple of boards (remember which channels they are) and send me a couple of close-up pics to assess the damage.
You can find pics of the machine and boards at Flikr
https://www.flickr.com/photos/33383402@N06/albums/72157658219628751/show
Sorry about the length of the post, but someone who is struggling with a repair might find something in it to help them. It was certainly well beyond my limited electronics knowledge to sort out.
Al