help with gates preamp I can´t sort out

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

rafafredd

Well-known member
Joined
Jun 3, 2004
Messages
2,409
Location
Rio, Brazil
I have four of these solidstate preamp boards by gates that I want to use for EQ gain make up.

So, I started tracing it by eye and hand just to figure out pinout. I thought it would be easy, but after some hours of thinking and not getting anywhere, I decided to trace it on paper to really see what was going on.

After doing so, I still can´t figure out the pinouts for my surprise. It´s a strange circuit. And to make things wrose, I can´t get hold of any info about the transistors used in the thing.I do not have any transistors pinouts, and I can´t figure out these, specially he the Q2 transistor. I thought it would be really easy to figure out the transistor pinouts as soon as I had it all traced, but again, some things in this circuit is over my head.

So, I was expecting some of you really experienced guys could take a look and maybe that´s something really easy for you. Any hint? I for myself was beaten by this four transistor boards:

reversed-1.gif


As you see, I used IEC to represent transistors. Q1 and Q2 are TZ1218. No info online as far as I could search... The output pair, Q1 and Q2 don´t even have any label on it.

So, looking at this, I can´t say much more other than obvious pinouts, like input (A) and output (J) etc... But B is a big ??

Any inputs welcome. If this is a usable ciruit after all, and if I think it sounds nice ater trying it, I may make another pcb for this, because the original layout doens´t look that good. I mean, if these are really DIY worth, because it would be so easy, with only four transistors. Cpuld work well as a line driver, low mpedance transformer driver, buffers or make-up stage for passive EQs.

About the reversing, I´ve checked 10 times, and that´s really what I have here.

BTW, CR1 and CR2 are MZ2361 wich I also can´t get any info about.
 
Model # or picture? I bet several of us have the manuals, since they are very common.
 
Not going to help you because you sniped the Sescom output Transformers from me on ebay..... damn you! :twisted: :razz: :razz: :razz: :razz: :grin: :cool:
 
Plain old NPN cascade with complementary CF output.

Does need a DC path from A to B. I betcha there was a transformer. Might tolerate 47K DCR here. Cap-coupling these things tends to lead to subsonic instability. Never saw that double-C decoupling in Q2 emitter. Might make cap-coupling into Q1 base simple or unstable.

Not bothering with incidental parts or factory part numbers:

35ioxg2.gif


Without knowing the supply voltage or transistor types I got a plausable result, V(DC(out)) ~~ 1/2 Vcc, so it aint fussy.
 
Now I see what was getting me all confused. This is the reference of the cold side of the secondary winding of the nput transformer... Because of cascade.

PRR, you are the master! Thanks for this! I will get some pictures of the boards, and I´ll get these racked with my new passive eq I´m working on. I just hoped I could get some connectors for these boards, but if I can´t, direct solder I will put on those these. I´m not taking boards out anyway!
 
> line driver, low mpedance transformer driver, buffers

No. {EDIT: no, no.... see below....}

Q3 Q4 buffer the load from that 100K resistor. They also have 10K in series with each base (an odd detail).

Unknown small transistors have Hfe of 50 to 600. "100" is a safe guess.

If the load is 1K, Q3 Q4 transform this by Hfe to 100K. The immediate consequence is that when Q2 is driven to zero collector current, the load can only go halfway to the rail, from 11.4V to about 17V, 6V swing when an 11V swing is possible at no-load. By this logic, 1K is a silly low load for this amp. Now, you can select Q3 Q4 for higher Hfe and maybe get to 500R at half-swing. But we are wasting a lot of supply voltage, and who wants that?

Wihtout knowing the app, I'm guessing it drove 10K pots and mix resistors. Maybe L R and Solo, so maybe 1,670R worst-case, with spec-claims taken for one or two loads and pots half-up, over 3K load.

Against this theory, we have that gigantic 500uFd cap which could be sized for 50R load, and that 150R secondary output which doesn't seem to give any protection to an output which sucks at 1K-500R load. Or maybe these parts are uniform for both input and output modules, even if silly for the input.

Alternatively Q3 Q4 could be very high Hfe Darlingtons, but the 2-diode bias is wrong.

CR1 CR2 can be whatever Silicon diode is cheap. The goal is to match the Vbe of Q3 Q4, but there is no way to do that (other than use same-type diode-connected transistors), so just stick something in. Set R10 to zero; in fact I think this should NOT be a trim but a fixed resistor. Start with zero or 10 ohms.

Supply voltage is uncritical. In simulation, anything 20V to 100V biases correctly and it aint real sick down at 12V. Smoke-test with 15V.

Transistor type is uncritical. Use Silicon jellybeans.

Use a floating DVM to read the voltage on one 10 ohm resistor. 10 milliVolts (1mA) is a bit cool, 100mV (10mA) is wasteful and hot. There's no right answer. The high impedance at the 100K resistor will mask crossover distortion and make a mockery of "50mV is ideal" rules of thumb. If running too cool, add 100R or 300R 1K or 3K at R10. If you smoke-test at 15V but plan to run 40V (upper limit for cheap jellybeans), aim cool at 15V and re-check at 30V before you go 40V. Or change your R15 100R to 1K for tests: if the amp sucks big current R15 will sag and keep power below the smoke point.

If you must fiddle: find the largest R10 which won't pass much more than 100mV across (10mA through) the 10R resistors, then vary down from there.

But I think if you get all your diodes and transistors from one lot, or at least can stick to the same brand, a fixed R10 will be fine.

Your R11=10K makes sense only as "protection" (and R8 makes less sense). If the output is shorted and large drive is applied, Q2 pulls Q3 hard and Q3 melts. More modern amps do this better. The need is rare, but when Q3 blows the gig is over, so it is worth protecting. With today's tougher Silicon, just taking output at "J" may be enough protection up to 48V supply.

The 240pFd+13K at A-B is clearly input transformer termination, acts above 50KHz. This hints at about 10K nominal secondary impedance, perhaps 1:8 mike input transformer, total gain of 100. Assuming 5V nominal output overload gives 50mV input overload, very typical of older broadcast studio practives (and way inadequate for most modern pop/rock music studio work). 10K nominal source also agrees roughly with the very low 10uA current in Q1. If we assume older Silicon could be a bit leaky and low-Hfe, then it seems just right.

Q1's emitter impedance is around 3K, 4K with the emitter resistor. Q2 input will be near 1K*Hfe or 100K, so Q1 has voltage gain of 25. Q2 flows around 0.2mA, gain of 80 or less. Total forward gain about 2,000, NFB gives 12, around 40dB NFB. Q1 swing is negligible, Q2 has a heap of local NFB in the emitter resistor. Open loop THD near a couple percent all 2nd harmonic, closed-loop way below 0.1% max and likely 0.02% up to nominal level.

I think they are worth using. I don't think it is especially worth reproducing. The gain=12 can't easily be changed without screwing the DC balance, the output won't drive random heavy loads, there's parts in there for no reason except the art of reliable transistoring was young.

You could re-work most impedances 10 times lower, cast-out idle parts, drive heavy loads. Best noise performance near a few K source, 1:2 mike input tranny or 10K:10K for general buffer-work.

AH-ha! Q2's emitter network is a R-C-R-C low-pass. If you replace transformer with a 20K resistor, use a 5uFd coupling cap to a low-Z source, we have a 3-pole filter with >180 deg phase shift, and Q1 is an amplifier closing the loop. Such things oscillate! Phono preamps were often this way, and could get very ugly below the audio band. However an ideal phase-shift oscillator needs gain of 27 to oscillate, and sloppy ones need more. The way Q2 loads Q1, gain is 25, maybe 50. So maybe it "can't" flutter, "by design".

But you know what I like better, after fooling with phono preamps: remove your C5, change C6 to 10,000uFd 6V. I taped up five 2,000uFd caps and I liked the sound, even though this hack was bigger than the rest of the board.
 
>> CR1 and CR2 are MZ2361 wich I also can´t get any info about.

> could be very high Hfe Darlingtons, but the 2-diode bias is wrong.


Oye, we are two fools. MZ2361 data is readily available, the "Stabistor Diode" series had 1 to 4 junctions in a package, this part is 1.24V to 1.38V at 10mA.
http://www.microsemi.com/datasheets/sa7-29.pdf

Which means each diode is 2 junctions, CR1+CR2 is four junctions, Q3 Q4 "must" be Darlingtons. (Otherwise we'd have ~~600mV across 10 ohms, 60mA, at say 15V, is about one Watt, and that's a lot since you did not mention heatsinks?)

OK, with new "Darlington" assumption, Hfe 1K-10K, then that 100K load on Q2 reflects out as 100R to 10R. It becomes perfectly reasonable to drive 600R loads. It is not silly to drive R18=150R into a 150:600 transformer to get huge line level from modest supply voltage.

So to build today, CR1+CR4 become four silicon tooth-pick diodes. Or two same-lot Darlingtons diode-wired. With perfect matching, to get above 1mA-10mA idle we need another ~~220mV-320mV of bias, which is just about 2K-3K resistance, which is about what R9||R10 do.

You need "naked raw" Darlingtons. TIP120 and chums have base resistors which need re-design; anyway they are total over-kill even if they are super cheap.

So they ARE Darlingtons, they ARE magic diodes, it IS a low-impedance driver.

However, there is no short-protection. Q3 Q4 can pass an entire Ampere, but not for long. Known short-proof loads are OK, but any unknown loads (the Outside World) must be fed from point J, past the 150R resistor. That will keep current down to tolerable levels.

I'll leave the above analysis in case someone wants a 3K load driver.
 
thanks, PRR, another lesson :thumb:

CR1 CR2 can be whatever Silicon diode is cheap. The goal is to match the Vbe of Q3 Q4, but there is no way to do that (other than use same-type diode-connected transistors), so just stick something in
This made me always think why nobody does it the ideal-way instead of the compromise route. Is the improvement from BE-diode connected transistors too low or is it budget?
 
12x means 21dB. perfect for Pultec makeup. Maybe passive EQ makeup was the original intended use on old gates boards and systems. Or compressor makeup, who knows...

Can´t use it for input buffer thought. I need unity voltage gain there. To put a 20dB divider before it should not be worth noisewise. So, I´ll make another buffer for the inuts and save the other two cards for another project, another EQ maybe.

If it sounds good, DIY version of this preamp card would be nice, with newer transistors. I also should recap the originals before trying it out since the old tantalums looks really bad. But I´ll try to leave the output lytics, since they sure looks good and tests good, and are called sprague atoms, whatever that means in guitar and hifi DIY circles out there.

PRR, I don´t want to be a pain in the a#$, but it would be nice to see the professor reworking the input stage for lower impedance, higher current, and lower noise... :green: It wold be a nice option for the DIY version.

Unfortunettly it seems we could not play with the feedback for lower and/or higher gain, for say, a mic pre, without messing the Q1 bias. But a 21dB gainblock is useful for sure.

Really, thanks for all! Really nice analysis!

Pics comes next!
 
http://img.photobucket.com/albums/v172/rafafreddy/DSC01973.jpg
http://img.photobucket.com/albums/v172/rafafreddy/DSC01975.jpg
http://img.photobucket.com/albums/v172/rafafreddy/DSC01980.jpg
http://img.photobucket.com/albums/v172/rafafreddy/DSC01983.jpg
http://img.photobucket.com/albums/v172/rafafreddy/DSC01985.jpg
 
final_reverse-1.gif


Works great in Circuit maker. I´m ready to fire it up.

At +24v supply, maximum output at 600 ohms is oly 16dBm. This is with R10 at 50%, 31mA through the diodes. I have to have a look at how R10 is set on the original boards. These are not here now. I´ll test the real thing at +24v and see how close circuit maker got :wink: Maybe these wer run on higher voltage, probably those transistors can stand up to 40-60v, but i´m not sure if I should try it. I wouldn´t want to fry the original impossible to replace transistors.
 
[quote author="rafafredd"]Works great in Circuit maker. I´m ready to fire it up.
...
At +24v supply, maximum output at 600 ohms is oly 16dBm. This is with R10 at 50%, 31mA through the diodes. [/quote]

I guess you mean 31uA through the diodes?

NB: in my version of circuitmaker the 4148 model is not accurate. Check things out with the 1N914 and/or 1N916 models if possible.
 
BIG difference. In Circuit Maker, four 914Bs gives just 10mA with trimpot R10 at 50%. I guess none of these models are perfect. 4001 also gives 10mA.

Voltage drop on 4148 is .68. Seems right???

914 and 4001 is just .48

:?
 
I think the 914 is closer to reality, although that (0.48V), in turn, sounds a bit low for the setup shown. But maybe not, at least not by much. The 400X parts should be lower Vf than the 914 for the same If, but which one is wrong I'm not sure. Probably both :grin:

Pease has some data in the back of his Troubleshooting Analog Circuits on forward voltages for various diodes, and I believe remarks something like "Nobody tells you this stuff!!" :grin:


With respect to an earlier post in this thread:

Horowitz and Hill in AoE, IIRC. have a brief discussion of the difference between diode-connected transistors and PN junction diodes, and the reasons for.
 
> At +24v supply, maximum output at 600 ohms is oly 16dBm.

Yeah. My counting-board sez +17dBm, full agreement.

The direct out would put 6dB more in 150 ohms. And 24V supply, it would be normal to use a 150:600 transformer to put serious level on a line. However, shorts happen, and that's not short-proof (except the small winding resistance). Logically that is what the 150R is for, but that will add 6dB loss to our 6dB gain, still just +16dBm.

It is not going to drive your cross-state telephone line cleanly.

It can soak a MOTU input, and what more could we ever ask?

> This is with R10 at 50%, 31mA

Does not matter what the bias is. Short the two Darling Bases together, it will still make the 40mW. The bias is so signals down at tenth-volt don't come out all distorted. For 10R emitter resistors, 5mA is a target. Going lower will probably-certainly crap-up soft sounds. Going higher may raise the THD/IM meter needle a pinch but will sound better to the ear. Huh.... 5mA is also essentially "class A" for 600 ohms and higher. So there's little reason to run hotter.

The thermal coupling of bias diodes to transistors is good to copy, though perhaps overly conservative here. The transistors won't run HOT. If they run near ambient, putting the diodes "next" to the transistors always worked fine. Under dynamic conditions there is some thermal lag and some say this matters. But making lag very low means putting the diodes ON the transistor chip, and this good idea was only invented last year. Ergo it must not be essential.

Odd to see the blob of goo technique. It gives good-enough thermal coupling to the PCB to gain some safe dissipation, though I don't know why they needed it. Maybe it is more mechanical than anything: I know those transistors (but not their numbers) and you can break them off the PCB on a long bad road-trip if you just stand them by the legs.
 
offhand I'm not sure about that particular Gates console, but they were fond of using 150:300 Daven ladders after preamps.
 
It is not going to drive your cross-state telephone line cleanly.

Shit! So, it's useless... :? :razz:

I love when you say such a thing, PRR!
But making lag very low means putting the diodes ON the transistor chip, and this good idea was only invented last year.

Wow, very interesting stuff. Would you have a part number?

Thanks for all the clarifications.
 

Latest posts

Back
Top