Help please!
I am having terminal issues with one of the channels on my CSX1 build - this seems to be destroying 2SJ28s at regular intervals.
This is only on 1 channel, the other is fine so far...
I had mica and grease separating my VFETs and the L shaped aluminium angle that they are mounted on. I have now replaced this with Kerafol 86/82. I have also taken precautions around the Gate and Source pins through the aluminium and the holding bolts to ensure there is no electrical short.
I have a LM317/LM337 based +/- supply with pass transistors for the main power. This has a cap and transistor arrangement to bring the V+/V- up slowly over about 8-10 seconds. The bias supplies are LM317 based and draw power from separate windings. These come up almost immediately.
I have managed to bias this up fine, and then on the next or subsequent turn on I suddenly find the problem channel has died.
The problem channel has:
32V on the Gate of the 2SJ28
23.66V on the Source of the 2SJ28
-5.7V on the connection between the Drains
-24V on the Source of the 2SK82
-32.7V on the Gate of the 2SK82
I had run this channel biased for ~60 minutes yesterday with no issue, biased at 1.2A with ~-5 to +10mV offset. The transistor case and aluminium angle next to it was touchable for 5-6 seconds so not particularly hot.
When I then test the 'blown(?)' 2SJ28s on a simple rig with 19V across, 0.7R in series with the drain to measure current and battery bias of 0-18V I get no current flowing through these VFETs even at ~0V bias.
If anyone has any ideas what could be going wrong then I am all ears. What could break my VFETs (3 so far
)? What could break them so they no longer conduct? I have lost 2 KF-33 and 1 KE-33 so I do not think it is batch related. Could it be mechanical, too much pressure? Could it be related to the heat of my soldering iron? Could I be getting some issue on switch off that is destroying the VFETs when the main regulated supply is switching off? Note the voltage of the bias supplies decrease slowly on turn off.
I have attached a picture of the problem channel, 2SJ28 is on the left. One bias supply is low to the left and the other (for the 2SK82) is higher on the right. The main power regulator is at the bottom of the shot.
ianc13
I am having terminal issues with one of the channels on my CSX1 build - this seems to be destroying 2SJ28s at regular intervals.
This is only on 1 channel, the other is fine so far...
I had mica and grease separating my VFETs and the L shaped aluminium angle that they are mounted on. I have now replaced this with Kerafol 86/82. I have also taken precautions around the Gate and Source pins through the aluminium and the holding bolts to ensure there is no electrical short.
I have a LM317/LM337 based +/- supply with pass transistors for the main power. This has a cap and transistor arrangement to bring the V+/V- up slowly over about 8-10 seconds. The bias supplies are LM317 based and draw power from separate windings. These come up almost immediately.
I have managed to bias this up fine, and then on the next or subsequent turn on I suddenly find the problem channel has died.
The problem channel has:
32V on the Gate of the 2SJ28
23.66V on the Source of the 2SJ28
-5.7V on the connection between the Drains
-24V on the Source of the 2SK82
-32.7V on the Gate of the 2SK82
I had run this channel biased for ~60 minutes yesterday with no issue, biased at 1.2A with ~-5 to +10mV offset. The transistor case and aluminium angle next to it was touchable for 5-6 seconds so not particularly hot.
When I then test the 'blown(?)' 2SJ28s on a simple rig with 19V across, 0.7R in series with the drain to measure current and battery bias of 0-18V I get no current flowing through these VFETs even at ~0V bias.
If anyone has any ideas what could be going wrong then I am all ears. What could break my VFETs (3 so far
)? What could break them so they no longer conduct? I have lost 2 KF-33 and 1 KE-33 so I do not think it is batch related. Could it be mechanical, too much pressure? Could it be related to the heat of my soldering iron? Could I be getting some issue on switch off that is destroying the VFETs when the main regulated supply is switching off? Note the voltage of the bias supplies decrease slowly on turn off.I have attached a picture of the problem channel, 2SJ28 is on the left. One bias supply is low to the left and the other (for the 2SK82) is higher on the right. The main power regulator is at the bottom of the shot.
ianc13
Attachments
If the bias went down from the bias, look for some thing is conducting from this circuit (Cap) to the other pole of the bias. The voltage on bias must be unchanged. Maybe the Cap is leaking and unstable. If you measure it? or change it.
The other parts looks fine and not leakage from the D to ground, you put your ohmeter...
Anybody more experience guy that help to avoid in burn more vfets in this cruel world.
The other parts looks fine and not leakage from the D to ground, you put your ohmeter...
Anybody more experience guy that help to avoid in burn more vfets in this cruel world.
the RIFA power cap are big enough to start a truck .. so i think the problem can be a defective one ( like i had with my Sitamp) or you can try lowering the value of the bleed resistor on them . burning very rare VFET what a nightmare
other thing , i now left my bias supply always on ( no power switch) so no surprise if my slowstart card turn on too fast..
other thing , i now left my bias supply always on ( no power switch) so no surprise if my slowstart card turn on too fast..
Last edited:
That's what I was thinking... and maybe you can put a big resistor in there which pulls 1.5 Ampere from the main supplies.
Last edited:
Thanks for the ideas.
The bias voltages decreases very slowly. There isn't much current flowing to discharge the caps in the bias circuits. I'd say 40 seconds to go from ~8V to ~2V after turn off. I will have a play with the main PSU disconnected to double check.
The main PSU discharges very quickly if the VFETs are conducting, the caps are big and bulky but only 10000uF each so I have 20000uF - 0.1R - 10000uF for each rail so half the size of the capacitor banks in Nelson's article.
On the other channel these discharge in a couple of seconds - i.e. ~24V to ~1-2V in 2 seconds or so.
I have never had a problem during actual running - I have had this channel biased and OK for 60 minutes with one VFET and maybe 30 minutes each with the other 2 VFETS with stable offset and current draw.
So it must be a start up of turn off issue
I have some IRFP240 and IRFP9240 at home so I am going to try and reconfigure the 24V regulated supplies to the scheme from the article with some different capacitors.
I haven't got any suitable Zeners in my parts bin though so I will have to pick those up online.
Is there anything that I can do to protect my VFETs from damage? Zener across Source/Drain or Gate/Drain like found internally in some MOSFETs or anything like that?
Thanks - ianc13
The bias voltages decreases very slowly. There isn't much current flowing to discharge the caps in the bias circuits. I'd say 40 seconds to go from ~8V to ~2V after turn off. I will have a play with the main PSU disconnected to double check.
The main PSU discharges very quickly if the VFETs are conducting, the caps are big and bulky but only 10000uF each so I have 20000uF - 0.1R - 10000uF for each rail so half the size of the capacitor banks in Nelson's article.
On the other channel these discharge in a couple of seconds - i.e. ~24V to ~1-2V in 2 seconds or so.
I have never had a problem during actual running - I have had this channel biased and OK for 60 minutes with one VFET and maybe 30 minutes each with the other 2 VFETS with stable offset and current draw.
So it must be a start up of turn off issue
I have some IRFP240 and IRFP9240 at home so I am going to try and reconfigure the 24V regulated supplies to the scheme from the article with some different capacitors.
I haven't got any suitable Zeners in my parts bin though so I will have to pick those up online.
Is there anything that I can do to protect my VFETs from damage? Zener across Source/Drain or Gate/Drain like found internally in some MOSFETs or anything like that?
Thanks - ianc13
Actually something else that I just thought of.
When I was testing the bad channel I was getting quite a lot of noise from my 300VA R-Core transformer in that channel.
I thought that I just had a noisy transformer or some DC on the line, but the other working channel is whisper quiet when biased at the same 1.2A.
I have generally found these Selectronic R-Cores to be really quiet for mechanical noise so could this indicate that I do have a bad cap before the 0.1R resistor which is drawing a lot of current and then behaving badly on power off?
I know that it cannot be after the 0.1R resistor because I am measuring the voltage across this to get an idea of the current draw.
When I was testing the bad channel I was getting quite a lot of noise from my 300VA R-Core transformer in that channel.
I thought that I just had a noisy transformer or some DC on the line, but the other working channel is whisper quiet when biased at the same 1.2A.
I have generally found these Selectronic R-Cores to be really quiet for mechanical noise so could this indicate that I do have a bad cap before the 0.1R resistor which is drawing a lot of current and then behaving badly on power off?
I know that it cannot be after the 0.1R resistor because I am measuring the voltage across this to get an idea of the current draw.
Hi dimkasta,
Yes I am insulating with fairly long plastic grommets from the underside. These are ~5mm deep after the flange and the aluminium angle is ~6.5mm thich so I also added a small amount (2-3mm) of heatshrink to very top of the bolts so that it is fully isolated from the chassis. The plastic grommets meet the heatshrink and squash it upwards when the bolts are tightened which should give complete isolation.
This does not affect where the top of the bolt contacts the VFET.
With the power supply and speaker connections disconnected I get infinite resistance to the chassis using my DMM for all the VFETs.
With the power supply connected I get a reading of ~220K Ohms for all VFETs on both channels which must be a function of the power supply itself. This is the same for both working and non-working channels.
This morning I sent a message but I don't know what happen.
This message was in relation to the problem of ianC13.
I was thinking about, and maybe the RC constant of the power supply is longer and the CAPs continues charged with full voltage and the bias goes out before, exposing the P channel transistor to more current, this could be worst if the Capacitors in parallel with the bias of this rail are with leakage and they going to discharge previously than the rail of the above mentioned component. Maybe you have to reinforce harder the bias and make some protection system that interrupter quick than the bias. I see this transistor is in the extreme of the L aluminium and the surface is not adequate in transmit the heat to the principal Heatsink. And, maybe the thermal inertia is longer, the case became hot for short periods, but with cumulative stress destroy the juncture.
This message was in relation to the problem of ianC13.
I was thinking about, and maybe the RC constant of the power supply is longer and the CAPs continues charged with full voltage and the bias goes out before, exposing the P channel transistor to more current, this could be worst if the Capacitors in parallel with the bias of this rail are with leakage and they going to discharge previously than the rail of the above mentioned component. Maybe you have to reinforce harder the bias and make some protection system that interrupter quick than the bias. I see this transistor is in the extreme of the L aluminium and the surface is not adequate in transmit the heat to the principal Heatsink. And, maybe the thermal inertia is longer, the case became hot for short periods, but with cumulative stress destroy the juncture.
The 2SJ28 is on the left when looking at the picture so has a lot of L angle. It is basically in the middle of the aluminium L angle so I don't think it is heat. I am running fairly conservatively 1.15-1.2 Amps and I can touch the heatsink around these 'middle' VFETs with no problem at all.
The heatsink around the 2SK82 on this channel (mirror on other channel) does get hot. I can still touch for >5 seconds so not burning - but I will resolve this because it could have offset issues as the amp warms up with 1 VFET running hotter than the other.
I will double check the rate of discharge of the bias again tomorrow, but I think it was quite a long time, of the order of 30-40 seconds.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.