Analysis of a classic MOSFET Hi-Fi topology under modified conditions

[tomsoon]

Hi Everyone,To expand the discussion with a concrete example, I would love to bring this schematic into the thread. This circuit represents a well-known, classic Hi-Fi amplifier design [June 1994 - Silicon Chip Online].As many of you know, the original Toshiba 2SK1530 and 2SJ201 FETs shown on this drawing are extremely rare today, and finding genuine ones on the market is nearly impossible [June 1994 - Silicon Chip Online].I would like to ask a few specific questions to the experts here regarding this exact topology, under a modified condition:What would happen if we built this specific Hi-Fi circuit with a reduced rail voltage of +/-55V instead of the original +/-70V, and replaced the rare Toshiba parts with dirt-cheap industrial pairs, such as the IRF540/9540 or IRF640/9640? Is it even worth trying?Under this +/-55V condition and with these dirt-cheap IRF replacements, what do you think the 10 kHz or 20 kHz square wave response would look like on a scope? How would the higher gate capacitances of these industrial devices affect the square wave tracking at these frequencies?In terms of sonic performance and classification, what would happen to this amplifier?Best regards,

 

[solderdude]

Do you see the input RF filter ?
220ohm 1.2nF = 600kHz -3dB when driven from 0 ohm and a bit lower when driven from 50 ohm or 600ohm out pre-amps.

A lower rail voltage means lower max output voltage swing.
Changing output devices and lower power supply voltages requires some other values (including stopper resistors on the MF) to change along with some other values.
Perhaps also the Miller cap (CQ9 to BQ3).

 

[tomsoon]

Szia solderdude,
Köszönöm a precíz mérnöki elemzést! Pontosan eltalálta a technikai részleteket a kapuütköző ellenállásokkal, az eltoláskövetéssel és a Miller-kapacitással (CQ9-től BQ3-ig) kapcsolatban [1994. június - Silicon Chip Online]. Pontosan ezek azok a paraméterek, amelyeket a hagyományos tervezési elvek előírnak, és amelyeket meg kell változtatnunk, amikor ezekre a nagy kapacitású ipari IRF eszközökre váltunk [1994. június - Silicon Chip Online].
De hogy ezt az elméleti forgatókönyvet egy kicsit tovább feszegessük a jó vita kedvéért:
Mi lenne, ha valaki megpróbálná ezt a közvetlen cserét anélkül, hogy megváltoztatná a panelen lévő ellenállások vagy Miller-kondenzátorok értékeit? A standard tankönyvi elméletek szerint az áramkörnek súlyos sávszélesség-csökkenéstől, lassú felfutási időtől vagy egyenesen instabilitástól kellene szenvednie.
Szakértői véleményed szerint alapvetően lehetetlen a fizika szempontjából, hogy egy ilyen gyári topológia tiszta, éles 10 kHz-es vagy 20 kHz-es négyszöghullám-választ tudjon fenntartani ezekkel az olcsó ipari alkatrészekkel, vagy létezhet egy rejtett áram- vagy elrendezési mechanizmus, amelyet a hagyományos szimulációk hajlamosak figyelmen kívül hagyni?
Üdvözlettel,

---

 

[SSS]

View attachment 552118Hi Everyone,To expand the discussion with a concrete example, I would love to bring this schematic into the thread. This circuit represents a well-known, classic Hi-Fi amplifier design [June 1994 - Silicon Chip Online].As many of you know, the original Toshiba 2SK1530 and 2SJ201 FETs shown on this drawing are extremely rare today, and finding genuine ones on the market is nearly impossible [June 1994 - Silicon Chip Online].I would like to ask a few specific questions to the experts here regarding this exact topology, under a modified condition:What would happen if we built this specific Hi-Fi circuit with a reduced rail voltage of +/-55V instead of the original +/-70V, and replaced the rare Toshiba parts with dirt-cheap industrial pairs, such as the IRF540/9540 or IRF640/9640? Is it even worth trying?Under this +/-55V condition and with these dirt-cheap IRF replacements, what do you think the 10 kHz or 20 kHz square wave response would look like on a scope? How would the higher gate capacitances of these industrial devices affect the square wave tracking at these frequencies?In terms of sonic performance and classification, what would happen to this amplifier?Best regards,

The lower Q1 current source needs to be adjusted when lowering rail voltage. For all other without measurement or simulation nothing can be said regarding the performance. It is a simple circuit which may work well but seems to be more a PA amp.

 

[pma]

Hungarian is a nice language.
Nem minden szarka farka tarka.

 

[voodooless]

Hungarian is a nice language.

Still AI generated, though

 

[tomsoon]

Instead of abstract simulations, here is the physical board layout from my CAD workstation. This amplifier was built exactly according to the stock component values from the schematic.
In fact, your intuition is spot on about the PA application. Several of these units were built with these dirt-cheap industrial IRF540 and IRF9540 devices specifically for a live band, and they were used heavily on stage for years without a single failure—all running with the original, unmodified stock component values, except for the IRF540/9540 replacements.

 

[tomsoon]

Still AI generated, though

Yes, and it will remain so, because I do not think in English and you will not understand.

 

[voodooless]

Yes, and it will remain so, because I do not think in English and you will not understand.

You also posted AI generated Hungarian…. Using it a simple translator is one thing, having it form arguments for you is quite another.

 

[pma]

Yes, and it will remain so, because I do not think in English and you will not understand.

But this forum rule is to post in English. And your post #3 is in Hungarian and is not translated to English. Check your PC settings.

 

[fpitas]

I'd start with a more modern design using available MOSFETs. There's nothing special about that design except the MOSFETs used. Most likely modern MOSFETs will require gate drivers, unless you find more exotic choices like SiC or GaN with low capacitance.

 

[tomsoon]

I'd start with a more modern design using available MOSFETs. There's nothing special about that design except the MOSFETs used. Most likely modern MOSFETs will require gate drivers, unless you find more exotic choices like SiC or GaN with low capacitance.

No exotic GaN FETs, no dedicated gate drivers, and no simulation tricks needed. Just standard, inexpensive industrial IRFs running with +/-55V rail voltage. This is the real-world 19 kHz square wave output on the workbench.

 

[tomsoon]

Sziasztok mindenkinek,
Hogy a végtelen elméleti viták helyett némi valós kontextust nyújtsunk, íme a fizikai elrendezési csomag. Ez az Eagle NYÁK-útvonaltervezése egy +/-56V sínfeszültséggel működő fokozathoz, szabványos, olcsó ipari IRF540 és IRF9540 eszközöket használva [1994. június - Silicon Chip Online].
Kérjük, vegye figyelembe, hogy ez csak egy kísérleti tesztpanel, nem a végleges verzió. Bár a NYÁK-elrendezés két pár FET-hez van tervezve, a végleges architektúra csak egyetlen párt használ. A valóságban ezt az egypáras konfigurációt egy 16 ohmos hangszóróterhelésre futtattam.
Nyugodtan építs egypáras verziót, végezz saját munkapad teszteket, és oszd meg itt a tényleges méréseidet. Ha bármilyen problémába ütközöl az építés során, jelezd, és örömmel segítek a javításban.
Üdvözlettel,

 

[fpitas]

No exotic GaN FETs, no dedicated gate drivers, and no simulation tricks needed. Just standard, inexpensive industrial IRFs running with +/-55V rail voltage. This is the real-world 19 kHz square wave output on the workbench.

Not bad for such a simple design. Be a little cautious with the high-frequency square waves. You can get shoot-through current, directly from one rail to the other, if the FETs aren't being turned OFF fast enough.

 

[DonH56]

Started to comment but quit because (a) audio amps were not my day job after college and (b) I don't speak (or read) Hungarian. As others have said, bias and compensation (due to changing gain) changes are likely to be needed when dropping the supply voltage significantly and changing output devices.

 

[fpitas]

Started to comment but quit because (a) audio amps were not my day job after college and (b) I don't speak (or read) Hungarian. As others have said, bias and compensation (due to changing gain) changes are likely to be needed when dropping the supply voltage significantly and changing output devices.

At this point I'm mainly thinking about lurkers who may want to learn something. Hence my warning about sufficient gate drive.

 

[DonH56]

At this point I'm mainly thinking about lurkers who may want to learn something. Hence my warning about sufficient gate drive.

I completely agree with you. I was thinking gate drive, and slew, could be an issue. As you know, many of my posts are pretty basic and below the level of the OP, but hopefully help those with less of a technical background. OTOH folk then correct me for not being detailed enough or endeavor to explain things to me at the same level since I don't publish my credentials in my sig. Hard to tell depth of knowledge on the 'net. I'd rather talk beneath someone's level and move up than go over their head and have them drop out. And hope I can exhibit the same grace to people talking (posting) to me.

No worries, and I'll echo other posts in saying good to see you around again.

 

[fpitas]

I completely agree with you. I was thinking gate drive, and slew, could be an issue. As you know, many of my posts are pretty basic and below the level of the OP, but hopefully help those with less of a technical background. OTOH folk then correct me for not being detailed enough or endeavor to explain things to me at the same level since I don't publish my credentials in my sig. Hard to tell depth of knowledge on the 'net. I'd rather talk beneath someone's level and move up than go over their head and have them drop out. And hope I can exhibit the same grace to people talking (posting) to me.

No worries, and I'll echo other posts in saying good to see you around again.

Thanks! Lots of great people here. It's good to be back.

 

[tomsoon]

Hi everyone,
To make things easier for anyone who seriously wants to build and test this configuration, I have attached the complete Eagle .brd file.
The layout was originally created using Eagle version 9.2.1, but as shown in the attached image, it works perfectly and is fully compatible with other versions as well. You don't need to spend time routing or designing the board from scratch. Just open the file directly in the Eagle software, and you are ready to produce the PCB.
Best regards,