IGBT Conversions for Forte Audio Amps?

Jump to Latest
#42
Oh, AWESOME. Thank you Jan, that is very helpful. I see that on the spec (data?) sheet: "Gate-Emitter Cut-off Voltage / Vge (off) / Vce=-10V, Ic=-1A"

Okay, so we know that the circuit is designed to work with an IGBT that requires a Vge(off) of -10V. Is there a way to translate that into a compatible Vth that I can look for in a MOSFET, or is this another one of those "It's not that simple, dummy!" things? 😄
 
#43
Well if the biasing system that is designed for 10V cutoff (or a bit higher), you don't want cutoff but a bit of bias current) modified for 3V-ish, that needs modification of the bias circuitry, the stuff around Q11.
There are no visible resistor values, that makes it kind of not easy. Do we know those values?

Jan
 
#44
Here is a general schematic for the Forte IGBT amps. A few things to note are it has no global feedback and uses Mikes variation of error correction like Cordell and Hawksford. The IGBTs had to be matched closely and torqued properly. While they had good points like easy drive (MOSFET) and high gain (Bipolar) they also had the bad, thermal runaway and oscillation. My brother has had one running for almost 30 years and it is still fine when I replaced a fuse last year. Most values are pretty close although R10 and R11 seem high. If I had to replace outputs I would use MOSFETs and make the gate resistors 475 Ohms. A scope and distortion measurement of some kind would be good. Finding a bias point would take some time.
 

Attachments

#45
Holy CRAP, Wayne, I can't even tell you how many times I have asked, and seen it asked by other people, for this schematic.. No response. The only Forte schematic I have ever seen available is for the Model 1. This is pure GOLD, my friend, and a hundred Forte owners thank you! Shoot, I may start selling this on the DIY Marketplace for $100 a pop.

Don't worry, I'll send you 10% of every sale. 😄

With the help of many friends here, I have narrowed down the MOSFET search to the attached table.

MOSFET transistors.png


These are just the P-channel, of course we would need to complimentary N-channel. Each has an adequate voltage rating, power dissipation, and all will tolerate a Vgs of between -20 - +20 volts. The IXTH8P50 is not so attractive with its 5 volt Vth, and I don't think we really need the 500V rating anyway. The IXTH36P10 is particularly interesting to me because of its high current rating, Vgs of -10V, and extremely low Vth of 25 MILLIvolts. (!) It also has a good form factor (TO-247-3) that should work well with the existing board/heatsink.

Does this make sense? Is there more information in this data that I'm overlooking or unaware of that would affect our decision?
 
#46
wayne said:
Here is a general schematic for the Forte IGBT amps. A few things to note are it has no global feedback and uses Mikes variation of error correction like Cordell and Hawksford. The IGBTs had to be matched closely and torqued properly. While they had good points like easy drive (MOSFET) and high gain (Bipolar) they also had the bad, thermal runaway and oscillation. My brother has had one running for almost 30 years and it is still fine when I replaced a fuse last year. Most values are pretty close although R10 and R11 seem high. If I had to replace outputs I would use MOSFETs and make the gate resistors 475 Ohms. A scope and distortion measurement of some kind would be good. Finding a bias point would take some time.
Click to expand...
One thing that I have found to be key to "maintaining" these older Forte amps is to remove the boards from the heat sinks and re-grease them with a minimum 6W/mK thermal compound, and, like you say, torque them correctly using the concave spring washers provided. This makes a big difference in stabilizing the transistors as we increase the bias voltage past 60 or 75 mV for best performance.

Don't know how you calculated that gate resistor, but that is very helpful (well, basically, because I wouldn't know how to calculate it myself). 🤔 Unfortunately, I'm not trained in the use of, nor do I own an oscilloscope, but maybe if we get the preliminary work done here, someone better than me will pick up the baton and test our results. No wins and losses here, only wins and lessons. 😉
 
#48
wayne said:
Here is a general schematic for the Forte IGBT amps. A few things to note are it has no global feedback and uses Mikes variation of error correction like Cordell and Hawksford. The IGBTs had to be matched closely and torqued properly. While they had good points like easy drive (MOSFET) and high gain (Bipolar) they also had the bad, thermal runaway and oscillation. My brother has had one running for almost 30 years and it is still fine when I replaced a fuse last year. Most values are pretty close although R10 and R11 seem high. If I had to replace outputs I would use MOSFETs and make the gate resistors 475 Ohms. A scope and distortion measurement of some kind would be good. Finding a bias point would take some time.
Click to expand...
Interestingly, this actual schematic is for the later Model 6a, with 75V rails and .33 ohm emitter resistors. The earlier Model 6 had .47 ohm emitter resistors and a second bias pot for DC offset, not present on the 6A.
 
#50
jan.didden said:
Well if the biasing system that is designed for 10V cutoff (or a bit higher), you don't want cutoff but a bit of bias current) modified for 3V-ish, that needs modification of the bias circuitry, the stuff around Q11.
There are no visible resistor values, that makes it kind of not easy. Do we know those values?

Jan
Click to expand...
Jan, I am sorry - I got so excited about the Forte 6A schematic that I'm ignoring your questions when you're trying to help. Here is the correct schematic that we are actually working with (THANK YOU WAYNE!):

Forte 6A Schematic.jpg


Which has resistor values!

It looks (to my uneducated eye) like bias voltage is being controlled at P1. Would we work with R18 and R19 to add resistance at R25-R28 and R37-R40, rather than change out those individual resistors to 475? Or, wait a minute. By messing with R14/15/18/19, am I not just affecting the rail voltage?

Hey, I never said I was smart. Just good lookin'. 🤠
 
#51
wayne said:
Here is a general schematic for the Forte IGBT amps.
Click to expand...
Wayne, could I have your permission to share this schematic with other Forte owners on the forum (just kidding about the Marketplace) while giving credit to you for the contribution? I have seen several (unanswered) posts from people looking for a Forte IGBT schematic that would benefit from this, but I want to confirm this is not proprietary or protected before sending it anywhere. Thank you.
 
#52
Nelson Pass said:
Yes, measure the Vgs you have.

😎
Click to expand...
Okay, Mr. Nelson - in utter silence 😉, I have measured the voltage from (what would be with MOSFETS) Gate to Source with the Variac set at 118V and got 2.3/2.4V (over six different IGBTs that I could reach without too much risk) at idle. With the positive lead on the gate, it's actually Negative 2.3/2.4V.

Am I right that this means I would want MOSFETs here with a maximum Vth of 2V?

As always, thank you for your input.
 
#54
Excellent news. That would indicate that our biggest challenge now is knowing which resistors and which values. 🤔

With Vgs at 2.4, and looking at the 6A schematic, would you say that Wayne's assessment of changing each individual gate resistor to 475 ohms would put us a little closer to home base?

Thank you
 
#55
R24 may need to be changed to get the bias set. The gate resistor change just prevents oscillations nothing to do with the bias.
I have been told that some very late model units when Threshold was in Camarillo used MOSFET drivers and bipolar outputs.
These changes would not be a minor undertaking without some test equipment.
 
#56
wayne said:
R24 may need to be changed to get the bias set. The gate resistor change just prevents oscillations nothing to do with the bias.
Click to expand...
This is the kind of stuff that I really appreciate from you guys. So in the change from IGBT to MOSFET, it sounds like bias and oscillation are (at least) two separate things we want to be aware of.

Perhaps P1 could be changed to a 5K bias pot to accommodate the difference? I like to change the original Bourns 3386P-1-501 (500 ohm, 7/8 turn, horizontal adjustment) pots to Bourns 3299/3296X (25 turn, vertical adjustment) pots anyway. They are so much easier to work with in the crowded environment of that little chassis. Installing a 5K version (502 instead of 501) of the 3296, and starting with it completely closed should keep us safe if we start up at 12V or so on the Variac.

So, just repeating what I think I've learned: we want to increase resistance at the bias circuit because the MOSFET may pass more current, and increase resistance at the gate to 470 ohms to control oscillation. If we then install some MOSFETs, at least we should have an amp that makes noise without melting down.

I'm still not completely sure if there is an intelligent way to choose transistors for a given circuit, or if the process of building an amplifier is choosing (or using) good components and then building the circuit around them. Nelson, I did read your "Arch Nemesis" paper and I can see how different transistors have a profound affect on THD in the same circuit. At least an input/output transformer circuit, I assume that's true for this circuit as well.

Wayne, you are right - we are probably going to need some test equipment and the knowledge of how to use it. One step at a time.
 
#58
Thanks, Nelson. Working on that now. Tektronix TDS 540C 500Mhz may be in the future. Nothing special, but it might be a start?

Still looking for some advice about choosing MOSFETs - is there any method, process, line of rational thinking? I'm sure we don't just grab a number out of a hat and work with that one this week. 😉
 
#59
I usually have good success with parts selection through the Mouser or Digikey websites.
If you make a short list of the parameters you need, like MOSFET, N channel, 200Vds(max), TO220 case, 20A Id(cont), in stock, etc, you can quickly narrow down the list of available parts.
At the end, you'd probably want to peruse the data sheets of likely candidates for things like Safe Operating Area.
Actually, selecting devices for your next project often looks just like work 😎

Jan
 
#60
Awesome, that's it. It's a data thing. I have an account with both of those suppliers, and that's exactly what I've been doing - it just seemed too "technical"..

I didn't know if there was some kind of audible, "audio" type transistors, as I have read many papers on capacitors and how they can (to some ears, anyway) affect the final sound of the circuit.

Perhaps the best logic here is to:
1. Acquire an oscilloscope. Is there anything special I should be looking for in an oscilloscope that would be required for audio? I don't want to invest that kind of money and find out I can't get the results I am looking for.
2. Learn to use the Oscilloscope for the purposes that will serve this project.
3. Install the 5K bias pots on the M5 circuit boards.
4. Order a selection of MOSFETs - say, 4 different models. I've only got this one pair of circuit boards, so all this soldering and solder sucking is going to get rough on them after a while.
5. Install each set of MOSFETs and test for ... frequency response? Distortion? Is there anything else you would normally look for in choosing a transistor?

I just want to say (one more time) that I consider myself very fortunate to have great minds like Jan Didden, Nelson Pass, and Wayne all still hanging out with me on this thread. Thank you for all your input!
 
You must log in or register to reply here.
Top Bottom