Ferrograph PPM driver circuit

Good afternoon DIYers.

I recently obtained a couple of PPM meters and driver cards for my Ferrograph Studio 8 tape recorder. They need some repair, and my Studio 8 manual doesn't have the circuit diagram (PPM was quite an unusual option), so I traced it out myself. I thought it might be of general interest to the DIY folks, but I'd also appreciate some help in understanding how it works:


(64KB).

I can see all the important parts of a PPM circuit: An input amplifier at bottom left, a full wave rectifier (IC2101), a peak detector (D2103, C2108), and a logarithmic amplifier (IC2102, TR2103, TR2106).

The peak detector must also include ballistics. The rectifier inverts, so the voltage at TP1 is more negative than the 'reference' at TP2. TR2102 charges C2108 rapidly to 0V through R2122, to register peaks quickly. The capacitor must then discharge slowly towards TP2. Is this through R2119? I don't get why R2119 is connected to the end of the rectifier circuit where it is.

The part that has me most confused is the logarithmic amplifier. I can see all the important parts of a log amp, but the way they're connected up doesn't look like anything I can find in my books. I'm also puzzled by the connections to IC2102. The input (from TP1) goes to the + input, and the - input is tied to TP2. There must be some negative feedback, but where?

(BTW, I forgot to put it on the diagram, but TR2103, TR2106 and TH2101 are all bundled together with thermally conductive grease).

TP2 appears to serve as a reference voltage for the circuit. This, and voltages 3V above and 13V below are created from the zener diodes D2105 and D2104. I'm not sure about the roles of transistors TR2107, TR2104, and TR2105. Are these part of the log amp, or are some of them involved in maintaining the voltage at TP2?

I hope my circuit diagram is accurate. I've checked it all quite carefully, but there could still be some mistakes that I've missed.

Thanks in advance for any insights that you can offer.

Steve.

> The capacitor must then discharge slowly towards TP2. Is this through R2119?

Mostly through R2123 82K. If IC2102 is working as an op-amp, both its inputs want to stay near TP2 voltage, so the 82K pulls C2108 toward that voltage.

> I don't get why R2119 is connected to the end of the rectifier circuit where it is.

IC2101b (uA739????) is the conventional 4-resistor one-opamp unity-gain differential amplifier, with TR2102 booster, except the top-right resistor has been corrupted into a 3-R network to give overall gain of 10.

> the logarithmic amplifier. ... The input (from TP1) goes to the + input, and the - input is tied to TP2. There must be some negative feedback, but where?

Output of IC2102 drives base of TR2103 with a (scaled-down) voltage. in response, TR2104 makes an exponential current. This is fed back to IC2101 input, in opposition to the current in R2123. Yes, this is the "+" input of the op-amp, but the transistor inverts so it comes out right.

M1 monitors the output voltage of IC2102, which is scaled-up from the voltage that TR2103 needs to cancel the R2123 current. The scale factor is about 12000/95 or about 120:1. Meter range is ~40dB. 40dB on a Silicon transistor is about 120mV, so the output of IC2102 must swing 14.4V. R'29, TH'01, R'34 scale this about 75% (the TH is likely 1K nominal) to ~10.8V. Effective resistance in series with M1 1mA is about 1K+1.1K+5.6K+(1.6K||4.9K) 9K. 10.8V/9K is somewhat close to 1mA, the right answer.

TR2106 is forced to flow a low current equal to the low end of the range to be logged. This ompensates the low-end temperature drift. It makes a lot more sense if R2132 is 6M8, not 6K8.... low-end current about (3V+13V)/7M9= ~2uA. Then if range is 40dB (IIRC it may be less) the top-end current is 200uA. 200uA*82K is 16.4V max on C2108 time-constant cap. Since the diff-amp after the rectifier has gain of 10, rectifier output for full scale is about 1.6V peak. The input transistors and IC2101a give a gain near 6.6. So input sensitivity is about 0.24V peak, ~0.17V RMS. This can be scaled to common levels with R2141.

You want a better puzzle? Figure this PPM:
click for big

Thanks PRR! It'll take me a little while to digest all that, but I'm sure it will be very helpful.

Did the way I'd drawn the circuit diagram make sense? It took me a while to work out the input amp and rectifier, but things got easier once some familiar shapes started to emerge. Being less familiar with the log amp end of things I had to guess a bit. I wondered whether there might be a more conventional way to draw the same thing.

[quote author="PRR"]
IC2101b (uA739????)
[/quote]
Yes, it uses a uA739. Is it unusual? This was the first time I'd seen one. Apparently it's an obsolete dual opamp in a 14 pin DIL package. Three pins per amplifier are for various compensation networks. I suspect a modern 8 pin dual could be substituted without too much difficulty.

It makes a lot more sense if R2132 is 6M8, not 6K8....

Click to expand...

Well, one of these boards definitely has a 6K8 there. It's mounted up on pins which suggests it might be a "Select on Test" part. The second board has 1M in the same position, which I hadn't spotted before now.

You want a better puzzle? Figure this PPM:

Click to expand...

One puzzle at a time, please! I think I've seen that one posted here before though. The string of diodes is a clever log approximation, IIRC.

Steve.

> Did the way I'd drawn the circuit diagram make sense?

You made that up just from looking at the PCB? Lot of work, apparently accurate (I may have doubts about some part values, but I did not see any "obvious" wrong connection), and a lot more readable than most factory docs!

I probably have not seen a 739 in 25 years. Aside from compensation frills, it has a resistor-loaded PNP pull-up output stage instead of the usual push-pull output. The rectifier stage apparently is so low-load that we don't need any push-pull power. The diff-amp that whacks the rectifier needs big peak current, and going the "wrong way" for the 739's output stage, so they add a buffer. Driving a cap is hard work anyway: maybe they liked it better this way than with a regular op-amp.

I've never seen a log-amp wired quite that way. First, of course, the whole thing hangs at a high positive DC voltage to confuse us (the floating meter is not confused). A divider between the opamp and the log transistors is common, but here we have two dividers. The whole mess biasing the right-hand (reference) log-transistor does not make sense at a glance. There may be a "better" way to draw it, but log-amps are always a mess anyway.

That's why I like the other design. It is simpler in the sense of less parts, mostly very obvious. Some of the junction offset corrections need a moment's thought. But the killer is the way it gets full-wave rectification in the one PNP.

[quote author="PRR"]
You made that up just from looking at the PCB?
[/quote]
Yes, it was quite a lot of work. Good fun, although I wouldn't want to have to do it too often. You certainly learn to appreciate a well drawn circuit when you've seen half a dozen inside-out and upside-down versions of the same thing.

I'd love to get hold of the original Ferrograph diagram to see how different that is. I think I have everything else from these tape recorders, except the PPM.

I probably have not seen a 739 in 25 years.

Click to expand...

The Ferrograph Studio 8 dates from about 1975, so that's a similar sort of age.

Steve.