Old Brazilian Amplifier Repair Attempt

[augustol]

So, I got myself a new project. Repairing an old 1974 brazilian amplifier (I'm also brazilian, by the way).

The model is a Gradiente Lab-75. I got it almost for free from an ebay-like marketplace we have here. The seller described as "Working perfectly, needs repair".

When I opened it up I immediately saw completely melted transistors and some loose wires. One of the channels (the one with melted transistors) had it's output disconnected. So, I started by replacing the original 1974 electrolytic caps and the melted transistors. Also replaced the output transistors for the original ones since the one in there were different (even though they were testing ok).

When I powered it on, I actually found the origin of the short. At first, I did some static tests I thought the origin of the short was one of the transistors. It actually wasn't. There was a film cap shorted. With the new transistors in, I got to this cap and replaced it for now with a ceramic just for now until I get the film cap to replace it.

The thing is, one of the transistors are still overheating (T6). I was studying the schematics and concluded the only place that could be having problems were the bias diodes. All the resistors around it are measuring ok. I did find an apparently bad resistor in the power line which might be making everything worse. I'll replace it.

The schematic I have in hand now looks to be a latter revision, it's not a good quality one and looks like it has some things different from what I'm seeing in the actual amp. Some components have two part numbers in the schematic (most of them obsolete parts) and the board only shows the older part numbers. The electrolytics had dates marked for early 1974. So I'm supposing this is and earlier model since this amp was released in 1974. So, it's kind of a perfect storm for a beginner, right?

Starting from the main strange thing I'm getting from the power supply: the schematic shows the transformer with two taps. One looks like 3v but might be 5V (yeah, the schematic is that bad) to power the indicator lamps which does measure 5V. The other tap is supposed to be 32V outputting (after a full bridge rectifier) 43/45/48V (the second digit is just a black blob of ink/pixels). I checked the diodes, they seem ok. The oscilloscope show a perfect straight line when I look at it. Almost no ripple. The cap after the bridge is a 2200uF and I got a new one to replace it, but when I was going to replace it I noticed the original one had four legs. I don't know yet if it's a dual capacitor (no indication on the body of the cap or the schematic) or if it's just for mechanical support. Haven't gone back there since since I see nothing that caught my attention in the oscilloscope.

The actual problem that led me to creating this post is: The strange diodes. There are 5 of them in each channel.

First is the OF128 (two in each channel) which I can't find any datasheets anywhere. There's one of them that seems not to be working but the other ones are measuring 0.7v drop across them, so I'm assuming they are just regular diodes.

The three other ones are a bit of a problem. The schematic indicates the part number BZX75 for the three of them. But the silk screen on the board says one is a BZX75C1V4 and one other is a BZX75C2V1. This ones I can find a datasheet for them and they are stabistors. C1V4 and C2V1 are the drops across the stabistors. And there is a third one with no markings on the board and the schematic only shows BZX75. And this one measure no drop across it in both channels. I'll have to remove and try to read those teeny tiny little letters in there.

So, the good thing is that all the drops are multiples of 0.7v, the bad thing is I'm a newbie and I don't know if can just put a couple of 1N400X diodes in series and it will be fine. I know that stabistors are more about the voltage drop than about the rectifying thing. Is there a better suited replacement diode for this application? Or any diode drop will be ok?

I'll see if I can attach the schematics I have here for you guys to take a look too.

[magic]

Actually, where do you see those diodes on the schematic you posted?

I see D3~D5, which are reverse biased for whatever reason. I don't think they are BZX75, more like BZX82 or BZX62?
Then I see D1~D2, drawn separately, presumably because they are mounted on the heatsink near output transistors? These are specified as ordinary 1N4148. They appear to be the actual diodes responsible for setting output bias.

[rsjsouza]

I agree with magic. In the schematic, the two 1N4148 are the bias diodes and are connected to the heatsink via wires "branco" or "br" and "preto" or "pr". The other three BZX62 diodes (on the schematic at least) seem to be part of overvoltage protection - they seem to be regular diodes. If in your amplifier they are BZX75, simply replace them with the same (if the soldering is original; otherwise it might have been "mexido" and someone changed it for whatever it was closer to them).

https://www.radiomuseum.org/tubes/tube_bzx62.html

(just like you, I can't find a reliable source of information about an "OF128")

Outside of this, does T6 heats up excessively when the loudspeaker is disconnected (S507)? If so, D1/D2 might be open (which could excessively bias its Base terminal), T8 is shorted, etc.

How are you measuring the resistors? Keep in mind that, if they are "in circuit", they might have their values changed.

 

[augustol]

Actually, where do you see those diodes on the schematic you posted?

I see D3~D5, which are reverse biased for whatever reason. I don't think they are BZX75, more like BZX82 or BZX62?
Then I see D1~D2, drawn separately, presumably because they are mounted on the heatsink near output transistors? These are specified as ordinary 1N4148. They appear to be the actual diodes responsible for setting output bias.

That's them. D1 through D5. The schematic shows D1 and D2 mounted externally, but in the amp they are actually in the same board as well as all of the other components. Only the output transistors are mounted on the external heatsink (T8 and T9). The heatsink also doesn't have any other mounting holes other than for the output transistors.

D1 and D2 are OF128's on the board wich I intend to replace with 1N4148's. D4 is marked on the board as BZX75C2V1 and D3 as BZX75C1V4. D5 is unmarked on the board. And they are all mounted in the direction indicated on the schematic in both channels.

I'm just reporting what I see on the board. But, now that you mentioned, the schematic does look like BZX 82 or 62.

I agree with magic. In the schematic, the two 1N4148 are the bias diodes and are connected to the heatsink via wires "branco" or "br" and "preto" or "pr". The other three BZX62 diodes (on the schematic at least) seem to be part of overvoltage protection - they seem to be regular diodes. If in your amplifier they are BZX75, simply replace them with the same (if the soldering is original; otherwise it might have been "mexido" and someone changed it for whatever it was closer to them).

https://www.radiomuseum.org/tubes/tube_bzx62.html

(just like you, I can't find a reliable source of information about an "OF128")

Outside of this, does T6 heats up excessively when the loudspeaker is disconnected (S507)? If so, D1/D2 might be open (which could excessively bias its Base terminal), T8 is shorted, etc.

How are you measuring the resistors? Keep in mind that, if they are "in circuit", they might have their values changed.

 

All of them do look original. The only changes I saw were a couple of caps and T6~T8. All the rest looks original.

I would like to replace the BZX's with BZX's but I can't find it locally. And by locally I mean country wide. I can only find it in stores in europe/usa.

I haven't connected any loudspeakers yet. I'm waiting for some 8R/100W resistors to come in so that I can make a dummy load to test it (it will take a while...).

I am testing this resistors in circuit. When I can't get a reasonable reading from the top, I find the pads and test it with the multimeter. I haven't removed any of them yet.

[dietert1]

The zeners provide output current limit, in conjunction with the 0.33 Ohm emitter resistors. Those low voltage zeners are multiple rectifier diodes inside. So you can replace the 2.1 V one with three, the 1.4 V one with two diodes in series. They put those so-called zeners to reduce parts count.

How old is that amplifier? You may have to renew the electrolytic capacitors.

Regards, Dieter

[magic]

Well, those "stabistors" aren't low voltage zeners, which is a different kind of animal, they are multiple diodes
That's what they look like, at any rate, given the specs: voltage drop of 0.58~0.67 at 1mA and thermal coefficient of -2mV/°C, per each "step" of nominal forward voltage.

BZX62 on the other hand are actual zeners, likely with different characteristics.

I think it makes sense that there are three diodes on the high side (two to cancel Vbe of T6,T8 and the third one to set maximum R42 voltage to ~0.7) and only two on the low side (one cancels T7 Vbe, one limits R43 voltage). Not sure what's the deal with D5, it looks like it should be a short in there

[augustol]

Well, those "stabistors" aren't low voltage zeners, which is a different kind of animal, they are multiple diodes
That's what they look like, at any rate, given the specs: voltage drop of 0.58~0.67 at 1mA and thermal coefficient of -2mV/°C, per each "step" of nominal forward voltage.

BZX62 on the other hand are actual zeners, likely with different characteristics.

I think it makes sense that there are three diodes on the high side (two to cancel Vbe of T6,T8 and the third one to set maximum R42 voltage to ~0.7) and only two on the low side (one cancels T7 Vbe, one limits R43 voltage). Not sure what's the deal with D5, it looks like it should be a short in there

I don't think those are zeners too. The multiple diode theory seems to make more sense.

It's funny you saying that D5 should be a short. The markings all over the board shows the symbol for the component and it's part number or value (no designator, though. but not important for this). D5 doesn't have it's part number marked and doesn't have the diode symbol (the bzx's have the diode symbol). It just has a straight line between the holes, exactly like some jumpers placed in other parts of the board. Might that be a mistake made by someone repairing it in the past (maybe induced by this same schematic)? The components that were replaced do look really old, from companies that I can't even find information on (brazilian manufacturers. as far as I know, brazil don't produce electronic component for many years now, maybe decades). Maybe they are not factory work?

But, about the drops... Supposing D3~D5 are just regular diodes, it makes sense the canceling of the Vbe's. But if you are canceling T6, T7 and T8 Vbe's, not canceling T9's wouldn't make the output a bit distorted because of this single Vbe not being canceled? Aren't push pull stages supposed to be symmetrical between low and high side? Also, aren't the BZX75's supposed to drop 1.4v and 2.1v? That kind of is a lot more than two diode drops.

How old is that amplifier? You may have to renew the electrolytic capacitors.

It's from 1974. Yes, I did replace the electrolytics. Some of them were in pretty bad shape haha

[magic]

This stage is not symmetric. T9 is an NPN wired as common emitter, not a PNP emitter follower.

That being said, the truth is somewhere in between. The output path contains not only T7 BE junction and R43, but also R47. It seems to be 13Ω and it conducts full base current of T9. I don't know what's the β of 2N3055 at 1~2A, but it probably is quite bad, which means that R47 may even have as much as almost 1V across it under peak load (and less when idle).

This contraption is reminiscent of so-called Baxandall diode, but not as accurate. The more proper components in this place would be a parallel combination of 100Ω and a diode - they conduct the same current as R41 and T9 base so their voltage drop would closely mimic T9 BE voltage. Besides helping the sanity of anyone trying to analyze the circuit, it supposedly reduces distortion too.

But R47 also has a second function - it reduces T7 gain by emitter degeneration. So if you mod the amplifier, it may end up needing a bodge capacitor between T7 base and collector to keep the output stage bottom half from parasitic oscillation.

edit
It may also provide some negative feedback against thermal runaway:
output current increases → T9 base current increases → R47 voltage increases → T7 base voltage decreases

So it may not be a good idea to mod this amp given the lack of proper (on-heatsink) compensation.
R47 voltage drop just needs to be considered in current limiting design calculations.

Also, aren't the BZX75's supposed to drop 1.4v and 2.1v? That kind of is a lot more than two diode drops.

Vf of a diode increases with current. The specs are in the datasheet: 0.57~0.67 at 1mA and about 0.7 at 10mA.
0.7V is a commonly used "rule of thumb" value of "one diode drop".

[augustol]

Ok, so... I got confused. Again...

Vf of a diode increases with current. The specs are in the datasheet: 0.57~0.67 at 1mA and about 0.7 at 10mA.
0.7V is a commonly used "rule of thumb" value of "one diode drop".

From the datasheet I've found the BZX75Cxxx are stabistors. I always have thought of those as series diodes. For example, if I put a BZXC1V4 on the circuit I'll get a 1.4V drop across it. Just like I would If had two diodes in series (~0.7v drop for each diode), right? In this case I have a D5 (with no markings on the board), D4 (BZXC1V4) and D3 (BZXC2V1 - 2,1V drop) in series. D3 and D4 will sum 3.5V drop across them, right?

I mean, I guess I'm confused on how does a Stabistor works. Am I right to think it works like multiple series diodes (but less sensitive to temperature changes)? Or it acts more like a zener, for example, like it has been mentioned here before.  Or is it a completely different kind of component?

This is the datasheet I'm using as reference:
https://pdf1.alldatasheet.com/datasheet-pdf/view/624533/NJSEMI/BZX75.html

[retiredfeline]

There's an entry on Wikipedia on it: https://en.wikipedia.org/wiki/Stabistor

Interesting device, learn something new every day.

[rsjsouza]

Yes, you are right to assume the stabistors are a bunch of Si diodes in series - the forward drop voltage is a multiple of Si PN junctions. However, since they all are inside the same packaging, the junctions are tightly coupled thermally.

Thanks for sending the datasheet for the BZX75 - their polarity on the circuit is indeed a bit strange, as they will be reverse biased during normal operation (that is why I imagined they would be for overvoltage protection). If you put them in reverse bias, they will not work as a Zener. Specifically for the BZX75, I can see from the datasheet actually shows a V_R of 10V. 

The schematic shows a quasi-complementary amplifier but, since you mentioned the bias circuit does not exist, I suspect the stability of the amplifier is very poor. I would perhaps re-design it to adapt to more modern parts if indeed the BZX75s are busted.

[magic]

As a general rule, Wikipedia is always full of shit
They have just enough correct information (of the sort that everybody knows anyway) to trick unsuspecting victims into trusting them.

These devices are specially designed for low-voltage stabilization applications requiring a guaranteed voltage over a wide current range and highly stable over temperature.

Yeah, right

- open the BZX75 datasheet on page 4 and compare the voltage vs current characteristic with a silicon diode like 1N4148 - very similar, but doubled/tripled/etc. Note that no claim is made of thermal stability and forward voltage is only specified at 25°C.
- open the BAS17 datasheet (link at nonsensopedia) on page 3 and look at "thermal coefficient" - it's specified -1.8mV/°C at 1mA, about what you would expect from a silicon diode.
- another datasheet - "approximately -2mV/°C".

So yes, these are more or less ordinary silicon diodes, particularly designed to have 0.7V drop at 10mA (which you will approximately get with any other diode too). Their voltage depends on current. It depends on temperature too. Hell, I'm pretty sure that even the exact thermal coefficient also varies slightly with current, like in any diode.

They are not Zener diodes. They are not precision devices. They are good enough for current limiting