my circuit to matching mosfet
Hi,
I use this circuit http://www.diyaudio.com/forums/attachment.php?s=&postid=552869&stamp=1105805193
100 Ohm for low power FETs such as IRF 610
10 Ohm for power FETs such as IRF 240
If the voltage is about 0mV, you have a perfect pair.
Both FETs thermal coupled and cooled.
I matched my sets of three for AX at 1,5A and 60 degrees C to 10mV.
greetings
Adam
Hi,
I use this circuit http://www.diyaudio.com/forums/attachment.php?s=&postid=552869&stamp=1105805193
100 Ohm for low power FETs such as IRF 610
10 Ohm for power FETs such as IRF 240
If the voltage is about 0mV, you have a perfect pair.
Both FETs thermal coupled and cooled.
I matched my sets of three for AX at 1,5A and 60 degrees C to 10mV.
greetings
Adam
Whatever you decide.
For parallel output devices with .47 to 1 ohm Source resistors,
less then .1 V Vgs variation is needed. At PL we match to .01V,
but that's because we can with no effort.
For input differential and other "signal level" pairs, it's mostly
a question of how much DC offset you want to put up with.
With input diff pairs, a lot of circuits, otherwise optimized and
with unity gain at DC, come down to whatever you want for
output DC offset. I regard 50 mV as perfectly adequate, and
usually trim if the value is greater.
Trimming is done in a half dozen ways, but an easy way is to
adjust the values of resistance coming off the Drains of the
input diff pairs.
The result - match as best you can and trim the rest.
For parallel output devices with .47 to 1 ohm Source resistors,
less then .1 V Vgs variation is needed. At PL we match to .01V,
but that's because we can with no effort.
For input differential and other "signal level" pairs, it's mostly
a question of how much DC offset you want to put up with.
With input diff pairs, a lot of circuits, otherwise optimized and
with unity gain at DC, come down to whatever you want for
output DC offset. I regard 50 mV as perfectly adequate, and
usually trim if the value is greater.
Trimming is done in a half dozen ways, but an easy way is to
adjust the values of resistance coming off the Drains of the
input diff pairs.
The result - match as best you can and trim the rest.
Thanks for the clarification Nelson. I've been matching to .01V, because it makes sense to me to match as close as possible so that the heat is evenly distributed and so the FET's done individually get too power hungry. Nice to have the grand master himself weigh in on this though
-Matthew K. Olson
-Matthew K. Olson
I thought it would be good to add
GRollins good advice to this old thread on same subject.
I agree with using some form of heatsink while testing power transistors.
And if you read previous posts here,
you will find several others also use heatsinking during matching of power devices.
GRollins good advice to this old thread on same subject.
I agree with using some form of heatsink while testing power transistors.
And if you read previous posts here,
you will find several others also use heatsinking during matching of power devices.
GRollins said:
Nelson,
I find this curious. Granted, you're prone to running things hotter than I am, but I tried running one "open air" and the readings across the resistor didn't stabilize--at least not before the device got way, way too hot to touch. Given that they were climbing, not falling, I didn't want to test the poor thing to destruction. Yes, at this point I've got a bunch of them on hand, but my frugal Scottish ancestry won't let me tear up a perfectly good device just to see if it stabilizes before it turns cherry red.
If I take the very first reading that I can see as the device begins to heat up, I find that it's on the order of 5-10mV higher than I get if I use a heatsink. I'm not comfortable with the variability of the 5-10 part. If it was more consistent then I could see going with it. Given that I can get a pretty stable reading in a couple of seconds with this aluminum plate I'm using, the time difference isn't all that bad. Where I do lose time is in the clamping. That costs me about five or ten seconds. Doesn't sound like much until you've done it a thousand times. (Literally.) Incidentally, I'm using a bulldog clip (like secretaries use to hold reports together) to hold the device to the plate. It's fast and provides consistent pressure.
One thing that occurs to me is that we may be using different meters. I use either a Fluke 8060 or and 8050 to do this sort of thing. This time I'm set up with one of my 8050s. I love Fluke meters, but there are times I wish that their sampling/calculation rate was a little faster. So my question is--what kind of meter are you using for matching?
I agree that the JFETs don't seem to be as sensitive as the IRF MOSFETs. It also helps that we're only doing 2W, as opposed to pushing 20-25W through a MOSFET. In principle, I could test the MOSFETs at some lower power, but I prefer to hit them with something closer to the real conditions they'd see in the circuit, so I toast them.
Grey
I find this curious. Granted, you're prone to running things hotter than I am, but I tried running one "open air" and the readings across the resistor didn't stabilize--at least not before the device got way, way too hot to touch. Given that they were climbing, not falling, I didn't want to test the poor thing to destruction. Yes, at this point I've got a bunch of them on hand, but my frugal Scottish ancestry won't let me tear up a perfectly good device just to see if it stabilizes before it turns cherry red.
If I take the very first reading that I can see as the device begins to heat up, I find that it's on the order of 5-10mV higher than I get if I use a heatsink. I'm not comfortable with the variability of the 5-10 part. If it was more consistent then I could see going with it. Given that I can get a pretty stable reading in a couple of seconds with this aluminum plate I'm using, the time difference isn't all that bad. Where I do lose time is in the clamping. That costs me about five or ten seconds. Doesn't sound like much until you've done it a thousand times. (Literally.) Incidentally, I'm using a bulldog clip (like secretaries use to hold reports together) to hold the device to the plate. It's fast and provides consistent pressure.
One thing that occurs to me is that we may be using different meters. I use either a Fluke 8060 or and 8050 to do this sort of thing. This time I'm set up with one of my 8050s. I love Fluke meters, but there are times I wish that their sampling/calculation rate was a little faster. So my question is--what kind of meter are you using for matching?
I agree that the JFETs don't seem to be as sensitive as the IRF MOSFETs. It also helps that we're only doing 2W, as opposed to pushing 20-25W through a MOSFET. In principle, I could test the MOSFETs at some lower power, but I prefer to hit them with something closer to the real conditions they'd see in the circuit, so I toast them.
Grey
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.