I want to build a mosfet amplifier (maybe the ESP P101) utilizing the transformer and heatsink assembly from a retired Tandberg 3012 integrated amp. The Tandberg was rated at 100W rms at 8 ohms / Ch.
The transformer secondary supplies about 57VDC +/- after smoothing to the rails. The original heatsink was quite robust with one pair of complementary mosfet outputs per channel. The original outputs were 2sj55 and 2sk175 rated at (ID) 8A and 125W ea. Tandberg seemed to think that these ratings were adequate for the 57V rails and functioned till the end without failure. Rod Elliot (ESP) seems to take a much more conservative approach with these higher rails (57V) and strongly suggests doubling up on the complementary pairs. As you can see in the pic the Tandberg heatsink assy. provides only for 2 TO-3 outputs per channel.
ESP P101 amplifier.
Rod recommends the Exicon ECX10N20 and ECX10P20 for the low power output. For my application (at the higher rail) I am considering the Exicon double die TO-3 versions, ECF20N20 and ECF20P20. I have read the pros and cons about the double die versions and still wonder if there are valid points to not using them. If it’s on the internet, it must be valid!?
Can I basically use the double die versions to replace the dual complementary lateral outputs without any real change to the P101 driver circuit at +/- 57VDC.
I emailed Rod and did not get a response.
OR… should I take the chance and use the lower power rated Exicons and justify it by what Tandberg used for power rated outputs. Seems to me that if this amp is used well below its rated output (like in my apartment), I should be able to get away with using the lower rated lateral outputs.
The transformer secondary supplies about 57VDC +/- after smoothing to the rails. The original heatsink was quite robust with one pair of complementary mosfet outputs per channel. The original outputs were 2sj55 and 2sk175 rated at (ID) 8A and 125W ea. Tandberg seemed to think that these ratings were adequate for the 57V rails and functioned till the end without failure. Rod Elliot (ESP) seems to take a much more conservative approach with these higher rails (57V) and strongly suggests doubling up on the complementary pairs. As you can see in the pic the Tandberg heatsink assy. provides only for 2 TO-3 outputs per channel.
ESP P101 amplifier.
Rod recommends the Exicon ECX10N20 and ECX10P20 for the low power output. For my application (at the higher rail) I am considering the Exicon double die TO-3 versions, ECF20N20 and ECF20P20. I have read the pros and cons about the double die versions and still wonder if there are valid points to not using them. If it’s on the internet, it must be valid!?
Can I basically use the double die versions to replace the dual complementary lateral outputs without any real change to the P101 driver circuit at +/- 57VDC.
I emailed Rod and did not get a response.
OR… should I take the chance and use the lower power rated Exicons and justify it by what Tandberg used for power rated outputs. Seems to me that if this amp is used well below its rated output (like in my apartment), I should be able to get away with using the lower rated lateral outputs.
Attachments
This seems to be non-sense for sound. Also, double-die Mosfet has the same capacitance as 2 single Mosfet in parallel. However, double die has the best matching between the 2 dies. For dissipation, double die has a bigger package than single and you simply need good heatsinks and good thermal contact.
Good luck
Fab
Twice as high rails for rated RMS power. Or power was limited by dissipation (sad, because user doesn’t know about). Or bad utilizing of supply voltage(what about best schemotechnic approaches).
The main cons is a relatively smaller thermal contact point, so try not to drive them hard.
Remember worst case for reactive load:
https://www.diyaudio.com/forums/solid-state/320968-soa-bjt-specialists.html#post5393240
Lateral FETs are much more robust than HEXFET or vertical DMOSFET like IRF240 and can sustain quite a lot of abuse unless the total dissipation is exceeded. Yet another factor lowering lifetime of a FET is gate-source overload which is not the case.
Second of all specs are about the same except the transconductance which may or may not contribute to distortions, biasing etc.
Thirdly, thermal interface material impedes heat transfer, a lot. So using two TO-3 instead of the one is advantageous indeed (but pricey). Metal can costs more than die itself.
Fourthly class B ideal efficiency is Pi/4 so you are kida safe with 4*125W (or 2*250w) at ideal heatsink trying to squeese the whole 100W.
Fifthly. Safe operating area or SOA. BJTs were really bad on this but not fets. Let's see.
Assuming no voltage sag of DC bus we'll get 57V/(8+1)=6A peak, where 8 Ohm is the load and 1 Ohm is Rds. Let's say inductive load and you are getting full voltage (i.e. 57 VDC) across the fet and max current (i.e. 6A) with Pi/4 phase shift between the former and the latter. This yields 342/2=171 W of instantaneous power. Safe again.
Despite #4 post double die package is the very same i.e. TO-3 so third assertion stays.
Same batch devices are usually indistinguishable but if you want to match fets look for threshold voltage first. +/-50 mV will do. That's said you get equal current sharing during transitions which is much more important than static.
Cheers
Second of all specs are about the same except the transconductance which may or may not contribute to distortions, biasing etc.
Thirdly, thermal interface material impedes heat transfer, a lot. So using two TO-3 instead of the one is advantageous indeed (but pricey). Metal can costs more than die itself.
Fourthly class B ideal efficiency is Pi/4 so you are kida safe with 4*125W (or 2*250w) at ideal heatsink trying to squeese the whole 100W.
Fifthly. Safe operating area or SOA. BJTs were really bad on this but not fets. Let's see.
Assuming no voltage sag of DC bus we'll get 57V/(8+1)=6A peak, where 8 Ohm is the load and 1 Ohm is Rds. Let's say inductive load and you are getting full voltage (i.e. 57 VDC) across the fet and max current (i.e. 6A) with Pi/4 phase shift between the former and the latter. This yields 342/2=171 W of instantaneous power. Safe again.
Despite #4 post double die package is the very same i.e. TO-3 so third assertion stays.
Same batch devices are usually indistinguishable but if you want to match fets look for threshold voltage first. +/-50 mV will do. That's said you get equal current sharing during transitions which is much more important than static.
Cheers
Last edited:
The devices mentioned by topic starter are in to-3 metal can package as well as the heatsink drilling is for. Plastic ones like 247/264 have, despite copper vs special tempco alloy in TO-3, exactly the same power rating. It well might be that die itself is the limiting factor in total power dissipation being, for instance, of a substantial thickness. But most pads are worse than metal so the higher the area the heat is transferred through the less thermal resistance will be.
Do not forget to reduce by half the value of the gate resistors (to have the same bandwidth as original ESP P101) since you have double die mosfet thus twice the input capacitance.
Also? Having built the P101 amp before and the slew rate is not as advertised but much slower - if it is important for you. This was confirmed by Elliott when communicating with him.
Good luck in your project.
Fab
FAB
"Do not forget to reduce by half the value of the gate resistors (to have the same bandwidth as original ESP P101) since you have double die mosfet thus twice the input capacitance."
I'll bring this up with Rod and see if he agrees with your suggestion. Unless there is some documentation from his sight on this topic. Thank you.
nigelwright7557
"Lateral mosfets are just silly money now, that's if you can find any.
I think profusion was the only company still selling none fake parts."
In the grand scheme of this project, I don't think that these Profusion TO-3 parts are that expensive considering the cost of 4 per channel and the over-all cost of everything.
"Do not forget to reduce by half the value of the gate resistors (to have the same bandwidth as original ESP P101) since you have double die mosfet thus twice the input capacitance."
I'll bring this up with Rod and see if he agrees with your suggestion. Unless there is some documentation from his sight on this topic. Thank you.
nigelwright7557
"Lateral mosfets are just silly money now, that's if you can find any.
I think profusion was the only company still selling none fake parts."
In the grand scheme of this project, I don't think that these Profusion TO-3 parts are that expensive considering the cost of 4 per channel and the over-all cost of everything.
I was finding after p+p they were about £20 each at Profusion.
You can get a lot of irfp240/9240 for that money and still get a great sounding amp.
The verticals need a different bias circuit to the laterals.
The laterals usually use a 1K pot but the verticals need a proper Vgs multiplier.
As I understand, the OP wants to use TO3 parts because the chassis already has provision for mounting them. That makes sense, rather than hacking the heatsink about, unless absolutely necessary.
The problem is that whilst Lateral mosfets are a sure-fire and simply applied device, the metal can varieties are close to un-obtanium presently. There was yet another thread here only week or so ago, moaning about general scarcity of lateral mosfets - check the recent threads if interested in the suggestions but I don't think you'll find much stock of genuine TO3 parts presently, let alone the double-die varieties.
The problem is that whilst Lateral mosfets are a sure-fire and simply applied device, the metal can varieties are close to un-obtanium presently. There was yet another thread here only week or so ago, moaning about general scarcity of lateral mosfets - check the recent threads if interested in the suggestions but I don't think you'll find much stock of genuine TO3 parts presently, let alone the double-die varieties.
Lateral MOSFETs | Profusion
Exicon T0-3 types in stock (not many), new, genuine and you can have them tomorrow. Nigel is right the double dies versions will be about £20 each after tax & delivery.
Exicon T0-3 types in stock (not many), new, genuine and you can have them tomorrow. Nigel is right the double dies versions will be about £20 each after tax & delivery.
