I have acquired a schematic for this amplifier, but I don't want to post it here due to copyright. I was wondering if anyone would be willing to shoot me an email, take a look at the schematic, and suggest which other components (besides the obviously fried ones) to check/replace. I don't want to put in ~$80 worth of transistors (what I found so far) only to have them fry again. My email is [my username] @gmail.com. Thanks in advance.
xplod1236 said:
When I build my amps I always fire them up without the output transistors connected.
I connect the output back into the LTP and then fire it up and check voltages. I always check the bias voltage is set to a minimum at this point.
Only if this is OK do I consider reconnecting the output transistors.
If it fires up all and well, I then set the bias using a sig gen and a scope.
That is a good tip. If the driver transistors have emitter resistors going to output (assuming a darlington type output here) then just leaving out output transistors is a good test. A series lightbulb in line with the mains is also a good idea.
Are the outputs blown on this amp? How many and which type? $80 in transistors sounds like a lot even for a big PA amp.
Are the outputs blown on this amp? How many and which type? $80 in transistors sounds like a lot even for a big PA amp.
megajocke said:
I usually find if the output transistors are blown then they go short across collector-emitter or drain-source.
A lamp across the input is a good idea.
I sometimes put in lower value fuses in the hope they will go before the output transistors. This is a juggling act, as too low a fuse and the switch on surge will blow them and too high and the output transistor die before the fuse blows.
One out of 6 (in parallel) MJ21194 output transistors is shorted. I will be replacing all 6. There are also 5 MJL21193s and 5 MJL21194s, which look to be in series with the outputs, which are completely blown up (melted legs, holes in them). Each one of those is $4-5 at digikey, so the total comes out to ~80, with a few smaller parts which I didn't mention.
xplod1236 said:
Thats quite a lot of money, that is why I use cheap MOSFETS in my designs. £2 each.
I would also have a dig around to see what might have caused all this damage.
Has your speaker fried ?
Is there a short in the speaker cable ?
I would seriously take my earlier advice and dont switch on first with all the output transistors connected.
Jus ttake short careful steps and things should be OK.
nigelwright7557 said:
Another trick I sometimes try is after checking the driver board voltages is to not connect all the output transistors but just one set on each phase. Then if you do blow them up its not a full set !
Have you seen that the MJ(L)21195/MJ(L)21196 are a little bit cheaper? They are a slightly newer, slightly better performance part. Almost identical, I mailed Onsemi once on the difference. I think it was that the 95/96 has a bit better gain linearity due to a new process.
Does it use both MJs and MJLs? Melted holes and stuff? I suggest you check the board for vaporized traces then too 😛 Driver transistors, emitter resistors, base resistors and so on. VI limiter circuit resistors, transistors etc. if used.
These use some kind of Class G? Output transistors in series with (a) diode(s) supplying middle rail to the junction? Commutation diodes could be bad.
edit: found it. Onsemi technical support said:
"The MJL21195 is considered a new generation of Perforated Emitter designs, better gain linearity and VBE(on) distribution but other than that parts are very similar. " [to MJL21193]
Does it use both MJs and MJLs? Melted holes and stuff? I suggest you check the board for vaporized traces then too 😛 Driver transistors, emitter resistors, base resistors and so on. VI limiter circuit resistors, transistors etc. if used.
These use some kind of Class G? Output transistors in series with (a) diode(s) supplying middle rail to the junction? Commutation diodes could be bad.
edit: found it. Onsemi technical support said:
"The MJL21195 is considered a new generation of Perforated Emitter designs, better gain linearity and VBE(on) distribution but other than that parts are very similar. " [to MJL21193]
Judging by the amount of power transistors, this is either a class G or class H amplifier and it's a bit harder to repair than common ones, particularly considering that all the rail switches (MJLs) seem to be destroyed.
I'm curious about the schematic of this newer model as I only have experience with the older AB 1100A (which uses MJ1502x stuff and +100/+50V/-50V/-100V rails and allows the rails to drop to 0V). Could you send it to me? evapowah at hotmail dot com
I'm curious about the schematic of this newer model as I only have experience with the older AB 1100A (which uses MJ1502x stuff and +100/+50V/-50V/-100V rails and allows the rails to drop to 0V). Could you send it to me? evapowah at hotmail dot com
"cheap MOSFETS in my designs. £2 each."
£2 = $4, looks like the same price to me, except they won't handle the power that the TO-3 cased MJs will.
In any event, this is a repair, not a new build. Kind of hard to make those £2 MOSFETS fit in (pin-outs are different if nothing else).
Knowing the general rail switching design AB uses (including a 0V rail switch), I wouldn't try to fire it up without the outputs. Use a lightbulb in series with one side of the line cord, 75W~100W should be fine for 120V.
This amp will do 4KW bridged, work safe!
£2 = $4, looks like the same price to me, except they won't handle the power that the TO-3 cased MJs will.
In any event, this is a repair, not a new build. Kind of hard to make those £2 MOSFETS fit in (pin-outs are different if nothing else).
Knowing the general rail switching design AB uses (including a 0V rail switch), I wouldn't try to fire it up without the outputs. Use a lightbulb in series with one side of the line cord, 75W~100W should be fine for 120V.
This amp will do 4KW bridged, work safe!
djk, I find it quite interesting that the TO-3s are CHEAPER than the MJLs at digikey. For an amp that has its thermal proteciton set at 95 degrees C the TO3s should equal at least two MJLs... Do you have the service manual for the amp? The rail switches (switching, not tracking) are quite nice and simple with very few discrete components 🙂
The drivers in this amp are supplied from the highest rail and they have emitter resistors going to output. There shouldn't be any problems running without outputs actually. But don't load it of course!! A high frequency output wouldn't be a good idea either as the zobel loads the amp at high frequencies.
I'd suggest you start by checking the following components to start with:
Vbe multiplier - a bad connection, intermittent transistor, resistor or bad pot here could eassily make the output blow up. R42-R44. Q15 The components could also be destroyed because of a destroyed output stage.
The Vas and its cascodes should be checked too as R65 which was blown connects the output flyback diodes. If the output went outside the rails there could be damage. Unlikely though.
VI limiter: Q17-18 D9-10, R50-51, R55-56
Driver stage: Q19-20, R96-R97, R52-53, R94-95
Commutation diodes: D19-22 If one of them is shorted the rail switch circuit might blow up even without outputs installed. The resistor joining output device collectors should be removed if powering up the amp without outputs and rail switches so the small signal transistors won't have to drive this load.
After checking these this it should be possible to check if the channel works up to the driver stage without installed outputs. A series light bulb is a good idea.
What type of transistors are used as drivers (Q19-20) now? The board layout says 2SC3281, 2SA1302. Are C18 and C19 installed and if so which value are they? 2SC3281 and 2SA1302 should be replaceable by 2SC5200 and 2SA1943 (or FJL4215, FJL4315) bur maybe the amp has been redesigned to take MJLs instead as drivers. (in that case I guess C18 and C19 will be missing or have a lower value)
The rail switches are also 2SC3281 and 2SA1302 according to board layout. If their equivalents can be used it will be cheaper...
edit: Saw that the circuit diagram says MJLs are used as drivers and rail switches. I'd use MJL21195/96 as drivers at least - they have better SOA than Toshiba/Fairchilds. The rail switches aren't as critical though.
The drivers in this amp are supplied from the highest rail and they have emitter resistors going to output. There shouldn't be any problems running without outputs actually. But don't load it of course!! A high frequency output wouldn't be a good idea either as the zobel loads the amp at high frequencies.
I'd suggest you start by checking the following components to start with:
Vbe multiplier - a bad connection, intermittent transistor, resistor or bad pot here could eassily make the output blow up. R42-R44. Q15 The components could also be destroyed because of a destroyed output stage.
The Vas and its cascodes should be checked too as R65 which was blown connects the output flyback diodes. If the output went outside the rails there could be damage. Unlikely though.
VI limiter: Q17-18 D9-10, R50-51, R55-56
Driver stage: Q19-20, R96-R97, R52-53, R94-95
Commutation diodes: D19-22 If one of them is shorted the rail switch circuit might blow up even without outputs installed. The resistor joining output device collectors should be removed if powering up the amp without outputs and rail switches so the small signal transistors won't have to drive this load.
After checking these this it should be possible to check if the channel works up to the driver stage without installed outputs. A series light bulb is a good idea.
What type of transistors are used as drivers (Q19-20) now? The board layout says 2SC3281, 2SA1302. Are C18 and C19 installed and if so which value are they? 2SC3281 and 2SA1302 should be replaceable by 2SC5200 and 2SA1943 (or FJL4215, FJL4315) bur maybe the amp has been redesigned to take MJLs instead as drivers. (in that case I guess C18 and C19 will be missing or have a lower value)
The rail switches are also 2SC3281 and 2SA1302 according to board layout. If their equivalents can be used it will be cheaper...
edit: Saw that the circuit diagram says MJLs are used as drivers and rail switches. I'd use MJL21195/96 as drivers at least - they have better SOA than Toshiba/Fairchilds. The rail switches aren't as critical though.
Megajocke, the only documentation I have for the amp is what I have emailed you.
I found Q15 to be shorted, along with R44. R43, the bias pot is bad as well.
VI limiter: both Q17 and Q18 are shorted. I'll have to pull D9 and D10 out of the circuit, because I'm getting weird readings on those. R55 and R56 are open.
Driver stage: Q19 and Q20 were completely destroyed. They are MJL21193 and MJL21194. C18 and C19 are 22pF.
I can't figure out which resistor that would be. I have built a few amplifiers out of power opamps (simple/easy), but this is my first time working on a discrete amp, so I have quite a bit to learn.
Thanks once again for all the help.
I found Q15 to be shorted, along with R44. R43, the bias pot is bad as well.
VI limiter: both Q17 and Q18 are shorted. I'll have to pull D9 and D10 out of the circuit, because I'm getting weird readings on those. R55 and R56 are open.
Driver stage: Q19 and Q20 were completely destroyed. They are MJL21193 and MJL21194. C18 and C19 are 22pF.
I can't figure out which resistor that would be. I have built a few amplifiers out of power opamps (simple/easy), but this is my first time working on a discrete amp, so I have quite a bit to learn.
Thanks once again for all the help.
"The resistor joining output device collectors should be removed if powering up the amp without outputs and rail switches so the small signal transistors won't have to drive this load."
I can't see the reference designator for that resistor - it's the resistor between NPN and PNP collectors. If you remove it temporarily you should be able to try the amp without output transistors and rail switches installed and measure output DC offset and that the bias voltage for the outputs is correct. A signal can also be applied with unloaded output to check for correct operation. Just check the commutation diodes in the rail switch circuit before applying power.
I'd say everything to the right of the VAS is supsect and needs to be checked... 😀 Maybe the feedback network resistor too. In the rail switches all transistors, diodes and resistors should be checked. I suggest you check components in the DC fault detector too.
I can't see the reference designator for that resistor - it's the resistor between NPN and PNP collectors. If you remove it temporarily you should be able to try the amp without output transistors and rail switches installed and measure output DC offset and that the bias voltage for the outputs is correct. A signal can also be applied with unloaded output to check for correct operation. Just check the commutation diodes in the rail switch circuit before applying power.
I'd say everything to the right of the VAS is supsect and needs to be checked... 😀 Maybe the feedback network resistor too. In the rail switches all transistors, diodes and resistors should be checked. I suggest you check components in the DC fault detector too.
I found that resistor: it's R80 on the schematic on the last page.
How would I measure the bias, and what would I set it to? I do have another working channel, so maybe I could use that as a reference.
I do still have the 6 PNP outputs installed on the board. Could I apply power with no rail switches, or would I have to remove those outputs?
How would I measure the bias, and what would I set it to? I do have another working channel, so maybe I could use that as a reference.
I do still have the 6 PNP outputs installed on the board. Could I apply power with no rail switches, or would I have to remove those outputs?
Are Q26, 43, 35 and 45 removed? If you disconnect R64 and R83 it should be OK to power it with transistors of one polarity installed. How much destruction is there in the rail switch circuits?
I noticed a connection that is missing on the schematic : Q26, 43, 35 and 45 emitters must be connected to each other but the line connecting 35 and 45 with the output collectors is missing.
The bias generator can be checked by making sure the bias voltage between driver transistor emitters (output bases) is adjustable. If it can't be brought below 0.7V or so there will be problems when installing output devices. The bias voltage must be turned down of course before starting it with output devices! Looks like some versions have a thermistor in the Vbe multiplier circuit. Did you check it (if installed)?
I'd say all components in the rail switch are suspect and need checking.
I noticed a connection that is missing on the schematic : Q26, 43, 35 and 45 emitters must be connected to each other but the line connecting 35 and 45 with the output collectors is missing.
The bias generator can be checked by making sure the bias voltage between driver transistor emitters (output bases) is adjustable. If it can't be brought below 0.7V or so there will be problems when installing output devices. The bias voltage must be turned down of course before starting it with output devices! Looks like some versions have a thermistor in the Vbe multiplier circuit. Did you check it (if installed)?
I'd say all components in the rail switch are suspect and need checking.
The 0.43ohm 2W base resistors of the rail switches are likely to be open (R107-R114).
The drivers of the rail switches are also likely to be damaged (Q21, Q22, Q35, Q36).
The 220ohm and 5.1ohm 1W resistors connected to the base and collector respectively of the rail switch driver transistors may be open or damaged too. The preceding stage is not likely to be damaged thanks to the 2K 1W resistors.
The rail clamping diodes (D19-D22) may be shorted.
The first power up seems difficult because there is no clamping to protect the output stage drivers from feeding the load directly when the rail switching is not working. It would be desirable to test the amplifier on the lower rails first.
The drivers of the rail switches are also likely to be damaged (Q21, Q22, Q35, Q36).
The 220ohm and 5.1ohm 1W resistors connected to the base and collector respectively of the rail switch driver transistors may be open or damaged too. The preceding stage is not likely to be damaged thanks to the 2K 1W resistors.
The rail clamping diodes (D19-D22) may be shorted.
The first power up seems difficult because there is no clamping to protect the output stage drivers from feeding the load directly when the rail switching is not working. It would be desirable to test the amplifier on the lower rails first.
What would be a direct replacement for an ST BDX54F transistor? This will be a replacement for Q21. I will be purchasing the components from mouser.com, so anything that they have in stock would be great.
"The first power up seems difficult because there is no clamping to protect the output stage drivers from feeding the load directly when the rail switching is not working. It would be desirable to test the amplifier on the lower rails first."
I find it strange they didn't include diodes from driver bases to the switched rails. (VAS is current limited so this should work fine) This way the drivers won't be able to drive output bases above their collectors. What's your favorite high speed diode for this voltage level and current?
I'm a bit worried about clipping too - if there is too much voltage drop over rail switches the drivers might try driving the output by themselves and blow up.
Other than this I like the design a lot. 😀 I did some simulations and it seems output stages with rails dropping to 0V makes a big difference in dissipation for reactive loads. It should be very useful on devices with bad high voltage SOA like the 2SC5200/2SA1943 or FJL4215/FJL4315.
I find it strange they didn't include diodes from driver bases to the switched rails. (VAS is current limited so this should work fine) This way the drivers won't be able to drive output bases above their collectors. What's your favorite high speed diode for this voltage level and current?
I'm a bit worried about clipping too - if there is too much voltage drop over rail switches the drivers might try driving the output by themselves and blow up.
Other than this I like the design a lot. 😀 I did some simulations and it seems output stages with rails dropping to 0V makes a big difference in dissipation for reactive loads. It should be very useful on devices with bad high voltage SOA like the 2SC5200/2SA1943 or FJL4215/FJL4315.
BAV21 / BAS21 are 250V small signal diodes, but I don't know if that is enough.
What are the actual rail voltages and at what mains input voltage?
That kind of output stage is very well suited to drive bass horns. Take a look at the predicted impedance plot for a long 40Hz folded bass horn, current phase is over 45 degrees lead/lag most of the time...
http://eva.eslamejor.com/HORNIMP3.gif
What are the actual rail voltages and at what mains input voltage?
That kind of output stage is very well suited to drive bass horns. Take a look at the predicted impedance plot for a long 40Hz folded bass horn, current phase is over 45 degrees lead/lag most of the time...
http://eva.eslamejor.com/HORNIMP3.gif
- Status
- Not open for further replies.