F5TV3 current hogging question
Hi,
I have been working on my F5Tv3 with four output pairs on and off for a while. Upon assembly things were looking good and the amp was biasing up well when unfortunately a slip of my multimeter probe seems to have shorted R5 and R27 on the store FE board and I heard a loud pop, some smoke and the 2A slo blo fuse popped immediately. Since I could not see any damage right away I put in another fuse and then there was another pop, some smoke and the fuse blew immediately. Diagnosing I found that the BJT's and the Jfets and other parts except the P2 pot(Nchannel side)on the FE board were fine. P2 replaced. All Toshiba Mosfets on the N channel and two Mosfets on the P channel output boards were blown. No visible damage to the Mosfets however. Most of the 50R gate resistors were swollen up and a black dot where they seem to have shorted to the board. The board looked fine. Anyway, I have replaced all the Mosfets and all the gate resistors on the N and P sides.
Now the amp fires up and biases well and I could reach my target bias fine. I got around 0.322V for the P channel and around 0.302V for the N channel. I then checked for bias voltage variations between devices on each side. For the P side, bias voltages were within 5mv of aech other. Great. However, for the N side there is a difference of 35mv between the bias with the lowest and the highest. All of the Toshiba Mosfets are bought from Nic who has super matched them. I too have checked the Vgs matching but at low current and voltage and they were within 10 mv of each other. The source resistors are Dale 1 ohm 1% 3W wirewounds. I did not replace the source resistors after the blowout since they looked fine. The diodes, source resistors and caps on the store output boards were not changed. I am puzzled by this variation. Could the source resistors be damaged on the N side even though they seem to be working fine on the other? Or what?
Advice would be greatly appreciated. Thanks.
Nash
Hi,
I have been working on my F5Tv3 with four output pairs on and off for a while. Upon assembly things were looking good and the amp was biasing up well when unfortunately a slip of my multimeter probe seems to have shorted R5 and R27 on the store FE board and I heard a loud pop, some smoke and the 2A slo blo fuse popped immediately. Since I could not see any damage right away I put in another fuse and then there was another pop, some smoke and the fuse blew immediately. Diagnosing I found that the BJT's and the Jfets and other parts except the P2 pot(Nchannel side)on the FE board were fine. P2 replaced. All Toshiba Mosfets on the N channel and two Mosfets on the P channel output boards were blown. No visible damage to the Mosfets however. Most of the 50R gate resistors were swollen up and a black dot where they seem to have shorted to the board. The board looked fine. Anyway, I have replaced all the Mosfets and all the gate resistors on the N and P sides.
Now the amp fires up and biases well and I could reach my target bias fine. I got around 0.322V for the P channel and around 0.302V for the N channel. I then checked for bias voltage variations between devices on each side. For the P side, bias voltages were within 5mv of aech other. Great. However, for the N side there is a difference of 35mv between the bias with the lowest and the highest. All of the Toshiba Mosfets are bought from Nic who has super matched them. I too have checked the Vgs matching but at low current and voltage and they were within 10 mv of each other. The source resistors are Dale 1 ohm 1% 3W wirewounds. I did not replace the source resistors after the blowout since they looked fine. The diodes, source resistors and caps on the store output boards were not changed. I am puzzled by this variation. Could the source resistors be damaged on the N side even though they seem to be working fine on the other? Or what?
Advice would be greatly appreciated. Thanks.
Nash
Thanks Vince. But that is not the issue. The issue concerns the large variation in bias voltage found between the four individual devices on the N side.
I just removed all of the source resistors. Since I dont have a CCS I used two 9V batteries and a 2.67K ohm resistor in series with each source resistor, measured the voltage across the two and calculated the resistance. Found some variation but not confident of my readings. Is it likely that the source resistors could be damaged because of the blowup?
Thanks.
Thanks for the suggestion.
The Vgs values he provided to me were 2.36, 2.36, 2.37, 2.36 for the 2SK1530 and 2.10, 2.11, 2.11, 2.11 for the 2SJ201. As noted in my post all the P channel 2SJ201's match very well.
Yes. Those are very good, but if matched them on 22v rails it is possible that when they are used with 42v rails that there will be some variation. As long as youmdp not have a. Variation greater than say 10%, you should be ok. 5% is better. Also, if you are using a lower value of Rs, matching tolerance must be higher.
the measuring of the Vgs should be done at the operating current of the amplifier.
Your supplier cannot be expected to know what your operating current is. All his customers will have a different operating current.
In effect he has supplied devices matched at the supplier's chosen current and voltage.
Checking your low ohms source resistors.
Put all you have in series. pass a current through them that is similar to your operating current.
Measure the voltage drop across each. Keep one of these as your "REF" so that you can find a match for one that gets damaged. Now select quad sets that have very similar voltage drops.
Your supplier cannot be expected to know what your operating current is. All his customers will have a different operating current.
In effect he has supplied devices matched at the supplier's chosen current and voltage.
Checking your low ohms source resistors.
Put all you have in series. pass a current through them that is similar to your operating current.
Measure the voltage drop across each. Keep one of these as your "REF" so that you can find a match for one that gets damaged. Now select quad sets that have very similar voltage drops.
Unfortunately I still have the problem.
I changed all of the source resistors on the N side with 1% resistors.
From front of amp to back the Nside bias readings when I originally posted two days ago were,
0.315 0.287 0.305 0.280V
After replacing the source resistors, removing all three caps on the output board and putting in a new R12 resistor, and reflowing joints even though they looked perfect, the readings are
0.315 0.243 0.306 0.263V
As you see the readings for the first and third device are virtually the same but the second and fourth have now changed a lot. Except for the diodes everything else on the output board is new.
Again, all devices on the P side read within 2mv of each other.
What am I missing? Please help.
Thanks.
I changed all of the source resistors on the N side with 1% resistors.
From front of amp to back the Nside bias readings when I originally posted two days ago were,
0.315 0.287 0.305 0.280V
After replacing the source resistors, removing all three caps on the output board and putting in a new R12 resistor, and reflowing joints even though they looked perfect, the readings are
0.315 0.243 0.306 0.263V
As you see the readings for the first and third device are virtually the same but the second and fourth have now changed a lot. Except for the diodes everything else on the output board is new.
Again, all devices on the P side read within 2mv of each other.
What am I missing? Please help.
Thanks.
Make yourself LM317 duda (read constant current source) then mesure V drop across resistor easy peasy
DIY Audio Projects - Hi-Fi Blog for DIY Audiophiles: Simple Low Resistance Measurement
Ordered LM317 two days ago. But I doubt this problem is connected with the source resistors. As mentioned the ones used are 1% which would result in a max difference of around 6mv at 0.3V bias.
The F5T V3 will work with out the diodes at 36V rail voltage?
The reason I would buy these PC boards so I can cascade the front end. I do not need such a high power 100w/ channel.
I have a 1KVA 2X30V transformer. I think these would work with out the diodes if I cascade the front.
Please let me know if I'm right about these.
Greetings
The reason I would buy these PC boards so I can cascade the front end. I do not need such a high power 100w/ channel.
I have a 1KVA 2X30V transformer. I think these would work with out the diodes if I cascade the front.
Please let me know if I'm right about these.
Greetings
Nic's mosfets were matched for Patrick's (EUVL) F5X at 16VDC, 2A, and 60C. Your operating points are very different from the matching conditions. Your results are not unusual at all. While the N and the P Toshiba mosfets are complementary, they don't response to the Vgs, current, and temperature variation the same way!!! I matched my own Toshiba's for my F5X, so I experienced that first hand.
If you really want tighter match, you can buy another matched quad, put them into your amp, measure, and select the best matches to your existing quad.
If you really want tighter match, you can buy another matched quad, put them into your amp, measure, and select the best matches to your existing quad.
When all possible causes bla bla bla
Just for fun twidle a litle bit the scew that hold mosfet in firm contact to heat sink
dirty,loose mosfet run hot and carracteristic are heat dependent
(we do not acept any responsability for stripped treads this is if U still use litle screw instead of bolt sunk in from opposite side and massive piece of bar to hold litle basterds in place like same nutters I know
)