My trusty Forte Model 5 and Model 6 are starting to show their age, and during a scheduled maintenance, I found an emitter resister completely blown (more than two VOLTS across the resistor when biasing), which, of course was caused by a bad IGBT. This led me to test all the output transistors (Very nicely matched within each amp, BTW) and I found one of the model 5 transistors to be bad as well!
Both were P-channel GT20D201, which (if you can find them) are only available as used pulls for around $20 each.
Which led me to start thinking about updating the entire board(s) with newer parts, and might not cost much more. I know the circuitry was designed around the IGBTs, and I really don't want to start rebuilding that. The latest version of Toshiba's IGBT is the GT50J121, which is faster, and has a higher current and voltage rating.
That's the background, here's the question(s): has anybody ever actually changed out a full set of IGBTs with a different model? Would the output circuitry be so carefully chosen that it would only sound right with the original transistors? Or, is there anything else you have learned from experience that would be a caution here?
Thanks for sharing your knowledge!
Both were P-channel GT20D201, which (if you can find them) are only available as used pulls for around $20 each.
Which led me to start thinking about updating the entire board(s) with newer parts, and might not cost much more. I know the circuitry was designed around the IGBTs, and I really don't want to start rebuilding that. The latest version of Toshiba's IGBT is the GT50J121, which is faster, and has a higher current and voltage rating.
That's the background, here's the question(s): has anybody ever actually changed out a full set of IGBTs with a different model? Would the output circuitry be so carefully chosen that it would only sound right with the original transistors? Or, is there anything else you have learned from experience that would be a caution here?
Thanks for sharing your knowledge!
Last edited:
BTW, it might be helpful to know that the form factor between the two transistors is the same (TO-3PL) and the pinout (Gate, Collector, Emitter) are the same as well.
i think these were the only complementary pair IGBTs ever made and they were not manufactured for very long. not sure you'll ever find suitable subs (i.e. good specs for audio use and kind of complements), other than NOS.
sorry ...
😟
sorry ...
😟
P-channel, not PNP.
n-channel IGBTs use an internal n-channel MOSFET and PNP bipolar structure integrated into a single device that is similar to an n-channel MOSFET
p-channel IGBTs use an internal p-channel MOSFET and NPN bipolar structure integrated into a single device that is similar to an p-channel MOSFET
Thus a PNP IGBT is ambiguous at best - it could be misinterpreted.
IGBTs are 4-layer devices so really they ought to be designated NPNP or PNPN, if you want to push the pendrantry (!)
Excellent, thank you Mark. I have a lot to learn in this field and comments like this one are very helpful. 👍
Thanks ML,
I kind of wondered why I couldn't seem to find any complimentary transistors to the GT50J121! 😕
There was a poster who wanted P-channel IGBTs, and did not respond later as to his progress.
It would be best to change the entire section to a matching set of conventional (bipolar or Mosfet) transistors, populated PCBs are sold, just select one matching your supply rails.
And do the replacement.
IGBTs have ageing issues, they have a limited life span, so a pulled one will have limited residual life. Also may have had excess heat applied during removal.
Not advised.
It would be best to change the entire section to a matching set of conventional (bipolar or Mosfet) transistors, populated PCBs are sold, just select one matching your supply rails.
And do the replacement.
IGBTs have ageing issues, they have a limited life span, so a pulled one will have limited residual life. Also may have had excess heat applied during removal.
Not advised.
Last edited:
Okay, this is interesting, and exactly what I was wondering.
Now, I'm going to admit to you guys that I am a copier technician by trade, not an electronic technician - so I have the aptitude, but not the education. If you feel it is worth your time to listen to what might seem like stupid questions, I would really appreciate it; but if you don't want to get dragged into another project, I totally understand. And one thing I WILL do is follow through with the results.
Although I have never been able to find a schematic or service manual for a Model 4, 5, or 6, I do have a schematic for a Model 1.
Looking at the output (power?) section of the board, it looks very similar.
This is a Model 5, which uses the same board, but has only three pairs of transistors instead of the usual five that are in a Model 1 or 6.
Questions: 1. Are IGBTs particularly susceptible to heat when installing/uninstalling, or is this a characteristic of all transistors? I'm asking because I removed one from the Model 5 board which I thought was good, but now does not seem to switch on when I apply 18VDC to the gate.
2. Since the heat sink is drilled and tapped for this board, it would be ideal to just replace the transistors with TO-3PL MOSFETS, as opposed to trying to install another populated board. Is the circuitry of this board specifically designed for IGBTs such that it will not sound right/work right with MOSFETS installed?
3. Is a populated board something I would find at Mouser/Digikey etc? What's the best search term - Output Section? Transistor Board?
Thank you for the education! 👍😊
Now, I'm going to admit to you guys that I am a copier technician by trade, not an electronic technician - so I have the aptitude, but not the education. If you feel it is worth your time to listen to what might seem like stupid questions, I would really appreciate it; but if you don't want to get dragged into another project, I totally understand. And one thing I WILL do is follow through with the results.
Although I have never been able to find a schematic or service manual for a Model 4, 5, or 6, I do have a schematic for a Model 1.
Looking at the output (power?) section of the board, it looks very similar.
This is a Model 5, which uses the same board, but has only three pairs of transistors instead of the usual five that are in a Model 1 or 6.
Questions: 1. Are IGBTs particularly susceptible to heat when installing/uninstalling, or is this a characteristic of all transistors? I'm asking because I removed one from the Model 5 board which I thought was good, but now does not seem to switch on when I apply 18VDC to the gate.
2. Since the heat sink is drilled and tapped for this board, it would be ideal to just replace the transistors with TO-3PL MOSFETS, as opposed to trying to install another populated board. Is the circuitry of this board specifically designed for IGBTs such that it will not sound right/work right with MOSFETS installed?
3. Is a populated board something I would find at Mouser/Digikey etc? What's the best search term - Output Section? Transistor Board?
Thank you for the education! 👍😊
Search for: Mosfet audio power amplifier kit
Read the wiki article, IGBT are mostly for drives, audio is a weird application for those.
Circuit may or may not work with Mosfets, and too much work.
The populated circuit boards start at $6 US here, much less than your hourly rate I think...
And like I said, find a replacement, and well, replace it.
Read the wiki article, IGBT are mostly for drives, audio is a weird application for those.
Circuit may or may not work with Mosfets, and too much work.
The populated circuit boards start at $6 US here, much less than your hourly rate I think...
And like I said, find a replacement, and well, replace it.
Last edited:
The variation in number of transistors seem to be a family, with the same basic circuit, and different power levels achieved by using different numbers of transistors in parallel.
So different models had the same design but different power levels.
And like I said, it is a weird application, only 1 factory in India makes those, they are the only people making IGBT based audio amps in the world!
Power levels are in kW!
That is PA / rock concert level...
So different models had the same design but different power levels.
And like I said, it is a weird application, only 1 factory in India makes those, they are the only people making IGBT based audio amps in the world!
Power levels are in kW!
That is PA / rock concert level...
The IGBT's were not my design, but I believe you should be able to fit complementary mosfets into that topology.
Wow, you guys are amazing. Thank you for taking time to contribute to my education. I downloaded the 40 page "IGBT Application Notes" from Toshiba, and the Wiki page is, of course, very helpful.
Vunce, that's actually a very good point you make - The Model 1 was NOT IGBTs, but used on a board with almost identical topology. That would indicate that the BJTs could be used on this board.
Naresh, the Forte 4, 5, 6, and 7 were a family of amplifiers designed by Michael Blasius in the late 80s, and all using the same circuit board (for both sides!). The 4, 6 and 7 shared a larger power supply, 30k uF capacitance, and five pairs of IGBTs per side. The 4 and 7 used the 58 volt tap from the 625VA transformer for 36VDC at the rails, and then was able to be biased up to class A (120-140mv) while still maintaining heat sink temperatures between 50-60 degrees C and producing 50wpc into 8 ohms. The 7s were monoblocks with more fin area per channel, so could be biased up a little more.
The 6 used the same components but used the 120 volt tap from the same transformer, making about 72VDC at the rails, and 200wpc into 8 ohms. With the higher voltage, it was only biased up to about 50mv, and is said to run class A up to the first 10 watts, then A/B.
The model 5 was kind of an economy version of the great circuit topology. It used the same boards, but a 400VA transformer, 15k uF capacitance and three pairs of IGBTs per side. The high tap from the transformer was around 90 volts for 56VDC rails and 100wpc into 8 ohms, at about half the current. But it made great sound, and in the day of $4500 Threshholds, it sold for a little under $1000!
Okay, one more dumb question: I see on the schematic that R19-R31 are emitter resistors, but I don't know what V14a-e and V15a-e are. Is this just designating the path to the bias pot (Variable Resistor) or is this an actual component?
Here's what I think we should do. I'm going to try to rescue an IGBT out of the Model 5 circuit board without damaging it, and replace the bad unit in the Model 6. Then, let's take the Model 5 boards and install a set of MOSFETs in place of the IGBTs. I have a lot less money invested in the Model 5, and it will be an interesting experiment, and probably useful information for anyone who comes across this thread in the future!
Vunce, you found 3281/1302 Bipolars in the Model 1, which would increase the chance that Bipolars would work with this topology. Nelson, you mentioned MOSFETs, which I tend to prefer the sound of but (I believe) don't have the current capability of the BJTs. Are there any particular MOSFETs that you have been working with that you would recommend?
Thank you all for contributing!
Vunce, that's actually a very good point you make - The Model 1 was NOT IGBTs, but used on a board with almost identical topology. That would indicate that the BJTs could be used on this board.
Naresh, the Forte 4, 5, 6, and 7 were a family of amplifiers designed by Michael Blasius in the late 80s, and all using the same circuit board (for both sides!). The 4, 6 and 7 shared a larger power supply, 30k uF capacitance, and five pairs of IGBTs per side. The 4 and 7 used the 58 volt tap from the 625VA transformer for 36VDC at the rails, and then was able to be biased up to class A (120-140mv) while still maintaining heat sink temperatures between 50-60 degrees C and producing 50wpc into 8 ohms. The 7s were monoblocks with more fin area per channel, so could be biased up a little more.
The 6 used the same components but used the 120 volt tap from the same transformer, making about 72VDC at the rails, and 200wpc into 8 ohms. With the higher voltage, it was only biased up to about 50mv, and is said to run class A up to the first 10 watts, then A/B.
The model 5 was kind of an economy version of the great circuit topology. It used the same boards, but a 400VA transformer, 15k uF capacitance and three pairs of IGBTs per side. The high tap from the transformer was around 90 volts for 56VDC rails and 100wpc into 8 ohms, at about half the current. But it made great sound, and in the day of $4500 Threshholds, it sold for a little under $1000!
Okay, one more dumb question: I see on the schematic that R19-R31 are emitter resistors, but I don't know what V14a-e and V15a-e are. Is this just designating the path to the bias pot (Variable Resistor) or is this an actual component?
Here's what I think we should do. I'm going to try to rescue an IGBT out of the Model 5 circuit board without damaging it, and replace the bad unit in the Model 6. Then, let's take the Model 5 boards and install a set of MOSFETs in place of the IGBTs. I have a lot less money invested in the Model 5, and it will be an interesting experiment, and probably useful information for anyone who comes across this thread in the future!
Vunce, you found 3281/1302 Bipolars in the Model 1, which would increase the chance that Bipolars would work with this topology. Nelson, you mentioned MOSFETs, which I tend to prefer the sound of but (I believe) don't have the current capability of the BJTs. Are there any particular MOSFETs that you have been working with that you would recommend?
Thank you all for contributing!
Actually, with a little more study, I realize that the Bipolar is a different topology in itself, using a base current to activate, where the IGBT and MOSFET both use gate voltage. Am I right that the MOSFET would be a better choice in this case?
Thanks
Thanks
No surprise, Nelson gave you a great suggestion.
You should take it 😉
"...I believe you should be able to fit complementary mosfets into that topology."
You should take it 😉
"...I believe you should be able to fit complementary mosfets into that topology."
IGBTs were more intended for relatively low speed switching applications rather than use in linear circuits, due to their low ON resistance.
True, but the Toshiba parts were specifically designed around audio. My own complaint with them was that they were a compound of Fet And BJT with no pin access to adjust relative biasing of the two components. This also complicated matching for parallel operation. In any case it is moot - Toshiba dropped the parts.
