Good news tonite.
Worked on the DIY Sony SIT Amp article (part 1) all day today. Maybe finish
tomorrow. Needs one pair 2SK82/2SJ28 per channel.
Time to get your parts from circuitdiy.com.
(Actually at last count they still have lots of pairs, but don't wait too long)
Attachments
Last edited:
Mr. Pass.
I've just read through your First Watt pages on your single-ended class A amplifier using your SIT device:
FIRST WATT SIT INTRO
I was wondering how you cope with having to use such a large value decoupling capacitor on the output? Is there a certain type of capacitor and configuration you recommend when using such a topology?
By my calculation for a 6 Ohm load and a -3dB point of 10Hz, a cap value of 2652uF would be required.
Thanks,
Darren
PS. Will your PASS-SIT-1 devices be available for purchase at some point (if they aren't already)?
I've just read through your First Watt pages on your single-ended class A amplifier using your SIT device:
FIRST WATT SIT INTRO
I was wondering how you cope with having to use such a large value decoupling capacitor on the output? Is there a certain type of capacitor and configuration you recommend when using such a topology?
By my calculation for a 6 Ohm load and a -3dB point of 10Hz, a cap value of 2652uF would be required.
Thanks,
Darren
PS. Will your PASS-SIT-1 devices be available for purchase at some point (if they aren't already)?
Regarding matching these parts, if one wants to parallel devices, do they need to be matched to each other as well as side to side? In other words, if one has a stash of 2SJ18's & 2SK60's, is it most critical to match the 18's & 60's as pairs, or is it most critical to match the devices that will be paralleled?
I'm trying to figure out how best to match these if one wants to build something LARGER than using one pair of devices, and am wondering how Nelson did it on the VFET amps he built for Sony? Unless the devices are very consistent, I'm guessing it would take a huge stash in order to get 16 closely matched pairs per channel if they were also matched to each other on each bank.
Am I making sense?
Finally, how close does the matching really need to be?
I am hoping to build 4 channels, using 8 pair per channel, chosen from a pool of about 50 pair. My friend is also planning on building 2 channels, using 4 or 6 pair per channel, from the same pool. Is this a realistic goal for us to achieve from a pool of 50 pair?
Any guidance would be appreciated.
Thanks,
Steve
I'm trying to figure out how best to match these if one wants to build something LARGER than using one pair of devices, and am wondering how Nelson did it on the VFET amps he built for Sony? Unless the devices are very consistent, I'm guessing it would take a huge stash in order to get 16 closely matched pairs per channel if they were also matched to each other on each bank.
Am I making sense?
Finally, how close does the matching really need to be?
I am hoping to build 4 channels, using 8 pair per channel, chosen from a pool of about 50 pair. My friend is also planning on building 2 channels, using 4 or 6 pair per channel, from the same pool. Is this a realistic goal for us to achieve from a pool of 50 pair?
Any guidance would be appreciated.
Thanks,
Steve
Last edited:
I appreciate the info thus far.
Do we know how Nelson matched these? Voltage and current settings would be most helpful. Also, perhaps even a test jig schematic if anything is different (or recommended) from testing normal fets. I do have the universal fet testing schematic that Nelson published long ago but not sure if it's applicable with these parts.
Since these are so rare, I thought it prudent to ask. It would be disappointing to accidentally damage any of them during testing.
Thanks!
Steve
Do we know how Nelson matched these? Voltage and current settings would be most helpful. Also, perhaps even a test jig schematic if anything is different (or recommended) from testing normal fets. I do have the universal fet testing schematic that Nelson published long ago but not sure if it's applicable with these parts.
Since these are so rare, I thought it prudent to ask. It would be disappointing to accidentally damage any of them during testing.
Thanks!
Steve
Last edited:
Acronman has been referring to the letter of the rank designation in their
offerings, for example I have been purchasing "E" designations.
This does give rise to some confusion which I think will be resolved shortly.
On the other hand, I am simply testing each part, and so should you. I heat
sink them and run them at 20 Vds and determine which Vgs value gives me
0.5 and 1.0 amp Ids.
So far for parallel operation I have been using values within 0.1 Vgs, and
they behave gracefully. Probably you could expand that tolerance to 0.2 V
or even higher without problems.
The only parallel case so far has been the big Sony VFET commemoratives,
the other examples being just single pairs of P and N which do not require
Vgs matching.
@ Apogee
Maybe this is response to your answer
Out of a 'random' 50 batch, N-ch generally show more uniform characteristics than P. In other words, you will get more matched groups of the N-ch devices than the P-ch, or larger matched groups for N-ch compared to P-ch.
However, it helps that they are pre-screened, using the rank number. Keep in mind that the rank number describes a fairly wide span of Vgs (2.5V) at 50V Vds. These may not be the conditions your amp will work at so it's better to chose some realistic values depending on you application. It is FAR easyer to get good matching amongst the same rank, since (in theory) the dispersion is reduced by a factor of 5.
If you do not intend to push the devices to their limits, then it's not absolutely crucial to have perfect current sharing between paralleled units, and larger tolerances in matching are possible. One easy way to achieve this is to desing-in a good safety factor - say, use 4 less matched units instead of 3 perfectly matched ones.
Finally, expect a number of odd-balls out of a large batch that will not be close to anything. I once did a matching (by curve tracer - tedious but really good matches!) of 16 2SK82, and got 3 x 3, 2 x 2, and 3 odd-balls, all of these were the same rank (3), so you can use this as an example.
However, it helps that they are pre-screened, using the rank number. Keep in mind that the rank number describes a fairly wide span of Vgs (2.5V) at 50V Vds. These may not be the conditions your amp will work at so it's better to chose some realistic values depending on you application. It is FAR easyer to get good matching amongst the same rank, since (in theory) the dispersion is reduced by a factor of 5.
If you do not intend to push the devices to their limits, then it's not absolutely crucial to have perfect current sharing between paralleled units, and larger tolerances in matching are possible. One easy way to achieve this is to desing-in a good safety factor - say, use 4 less matched units instead of 3 perfectly matched ones.
Finally, expect a number of odd-balls out of a large batch that will not be close to anything. I once did a matching (by curve tracer - tedious but really good matches!) of 16 2SK82, and got 3 x 3, 2 x 2, and 3 odd-balls, all of these were the same rank (3), so you can use this as an example.
By the way, many thanks for the Kubrick print. I'm getting a nice frame for it.
You're more than welcome. I'm really glad you like it.
I do no know if you saw my friend dsdjoy results….
http://www.diyaudio.com/forums/pass...2sk82-e-rank-2sj28-f-rank-11.html#post3855436
http://www.diyaudio.com/forums/pass...2sk82-e-rank-2sj28-f-rank-11.html#post3855436
I just finished matching my VFets Sony K60 (JG-34) and J18 (JF-24). I used 24.5 V regulated as Drain supply with a drain load resistor of 8 ohm/40w to monitor the voltage drop on this to get 500ma and 1 Ampere flow to the device.
K60 yeilds a Vgs of -7.8v @.5a and -4.99v @ 1a. other closely identical (Range .1-.6v difference)
J18 yeilds a vgs of +8.28 @ .5a and +5.91v @ 1a and others has close reading.
I know matching is only for parallel operation.Do I need to match the K60's Vgs with the J18's too, or this results will do?
K60 yeilds a Vgs of -7.8v @.5a and -4.99v @ 1a. other closely identical (Range .1-.6v difference)
J18 yeilds a vgs of +8.28 @ .5a and +5.91v @ 1a and others has close reading.
I know matching is only for parallel operation.Do I need to match the K60's Vgs with the J18's too, or this results will do?
Thanks....One more thing what will be the maximum difference of VGS for the P and the N?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.