Hi guys,
Siliconix have ditched my favorite matched dual JFET, the SST404. I knew it was too good to be true. A new mob, Calogic, appear to be continuing to run the fab and making parts, but the end result is that I can't find them in small quantities.
I'm chasing up a (reasonably) bulk lot of fifty SST404s, which should do me a goodly time, but this whole business has me frightened about using packaged matched duals.
So I'm looking at single JFETs, as there's a reasonable range of different parts that I can design with. I can hand-match singles, in say a SOT-23 package for Vgs at some arbitrary Id. But I'm concerned about how closely I'll have to couple the devices thermally to maintain that match to within reasonable tolerances.
What are people's experiences? If you match a pair, what do you need to do to ensure they stay that way over a reasonable operating range for an amp? Is locating them next to one another on a PCB adequate? Or do you need to bridge them thermally, using for example a bit of (horrendously expensive) gap pad and a little heat sink?
Cheers,
Suzy (sobbing in her beer)
Siliconix have ditched my favorite matched dual JFET, the SST404. I knew it was too good to be true. A new mob, Calogic, appear to be continuing to run the fab and making parts, but the end result is that I can't find them in small quantities.
I'm chasing up a (reasonably) bulk lot of fifty SST404s, which should do me a goodly time, but this whole business has me frightened about using packaged matched duals.
So I'm looking at single JFETs, as there's a reasonable range of different parts that I can design with. I can hand-match singles, in say a SOT-23 package for Vgs at some arbitrary Id. But I'm concerned about how closely I'll have to couple the devices thermally to maintain that match to within reasonable tolerances.
What are people's experiences? If you match a pair, what do you need to do to ensure they stay that way over a reasonable operating range for an amp? Is locating them next to one another on a PCB adequate? Or do you need to bridge them thermally, using for example a bit of (horrendously expensive) gap pad and a little heat sink?
Cheers,
Suzy (sobbing in her beer)
Hi Suzy
I have been trying similar things.
I still have some 2sj109 and 2sk389 but I find it difficult to buy more at "reasonable" prices. So I have been experimenting with 2x2sj74 and 2x2sk170 pairs. I began to experiment after noticing that the Borbelly amps and kits often used the pairs of fets instead of the dual fets.
I have been building two complementary symmetrical fet pre amps. The input stage is a complementary differential stage not unlike the Borbelly starter designs but without the folded cascode on the output stage. ( Although I might add a folded cascode later as I read that Borbelly and John Curl both seem to prefer that arrangement. ) The output stage jfets are connected at their drains to act as constant current sources to each other.
I expected some difficulty with dc on the output due to temperature effects on the fets. To my surprise once I have set the output dc to zero I am finding little or no dc on the output on the pre amps.
I did match the fets when building the pre amps. I had a 100 2sk170bl and 100 2sj74bl and used matched sets of each. I am using 2sk170bl and 2sj74bl on the input with 3ma drain current on each jfet. The output stage has 8ma drain current.
The temperature in the pre amp case does not vary much so maybe that is why I get no dc drift. However I think it is largely due to device matching and complementary symmetrical layout as moving the pre amp from room to room does not seem to have any effect. I have not found it necessary to connect the fets with heatsinks - although I did lay them out so that I could do so if necessary.
I am planning to try some fets in the input syage of a class A power amp next so I let you know what happens then.
Don
I have been trying similar things.
I still have some 2sj109 and 2sk389 but I find it difficult to buy more at "reasonable" prices. So I have been experimenting with 2x2sj74 and 2x2sk170 pairs. I began to experiment after noticing that the Borbelly amps and kits often used the pairs of fets instead of the dual fets.
I have been building two complementary symmetrical fet pre amps. The input stage is a complementary differential stage not unlike the Borbelly starter designs but without the folded cascode on the output stage. ( Although I might add a folded cascode later as I read that Borbelly and John Curl both seem to prefer that arrangement. ) The output stage jfets are connected at their drains to act as constant current sources to each other.
I expected some difficulty with dc on the output due to temperature effects on the fets. To my surprise once I have set the output dc to zero I am finding little or no dc on the output on the pre amps.
I did match the fets when building the pre amps. I had a 100 2sk170bl and 100 2sj74bl and used matched sets of each. I am using 2sk170bl and 2sj74bl on the input with 3ma drain current on each jfet. The output stage has 8ma drain current.
The temperature in the pre amp case does not vary much so maybe that is why I get no dc drift. However I think it is largely due to device matching and complementary symmetrical layout as moving the pre amp from room to room does not seem to have any effect. I have not found it necessary to connect the fets with heatsinks - although I did lay them out so that I could do so if necessary.
I am planning to try some fets in the input syage of a class A power amp next so I let you know what happens then.
Don
Hi
With the differential j-fet input and gnf, I find that matching of the devices is related to DC offset. In one of my "ugly amps", I very closely matched one channel, but deliberately matched the other not quite as close just to see the effect. There were a few more mV of offset, right about were I figured. I haven't used duel devices, they cost a lot more and are harder to find, as you know. There is more variety of singles anyway. If you choose to go with singles and hand match, you should get a decent quantity of them. There will be much greater difference in matching Vgs in j-fets than beta or Vbe in bipolars. My method is to sort them in general groups measuring at whatever Id you plan to operate them, and then from each group closely match them into pairs when you go to build the amp. Get some small sticker labels to keep the closely matched pairs together. I'm sure there are other methods, but this simple setup works for me.
Accuracy depends on how much time and effort you invest in it. As for SOT-23, I would think if they are mounted in the same place, the temp should be the same if you cascode the input j-fets so Pd is the same. I haven't experienced any real problem with this.
With the differential j-fet input and gnf, I find that matching of the devices is related to DC offset. In one of my "ugly amps", I very closely matched one channel, but deliberately matched the other not quite as close just to see the effect. There were a few more mV of offset, right about were I figured. I haven't used duel devices, they cost a lot more and are harder to find, as you know. There is more variety of singles anyway. If you choose to go with singles and hand match, you should get a decent quantity of them. There will be much greater difference in matching Vgs in j-fets than beta or Vbe in bipolars. My method is to sort them in general groups measuring at whatever Id you plan to operate them, and then from each group closely match them into pairs when you go to build the amp. Get some small sticker labels to keep the closely matched pairs together. I'm sure there are other methods, but this simple setup works for me.
Accuracy depends on how much time and effort you invest in it. As for SOT-23, I would think if they are mounted in the same place, the temp should be the same if you cascode the input j-fets so Pd is the same. I haven't experienced any real problem with this.Attachments
I have matched (sorted) hundreds of BF862's as they are used in my pocket Class A amps. I make the actual circuit that they will be used in on a small PCB fitted with all the resistors to simulate the in-situ measurement of the bias current across the drain resistor. Using the rubber eraser end of a pencil, I apply pressure to the SOT23 part for good electrical contact. The hard part is waiting for it to thermally stabilize before taking the reading. In a lot of 40 I can usually find 2 pairs that match currents better than 0.5% and 3 to 5 pairs that are within 1%. Sometimes one pair is within 0.2% - those are special and I set aside for my special top amps.
This thread has some good info:
Jfet matching jig
This thread has some good info:
Jfet matching jig
Last edited:
I have matched (sorted) hundreds of BF862's as they are used in my pocket Class A amps. I make the actual circuit that they will be used in on a small PCB fitted with all the resistors to simulate the in-situ measurement of the bias current across the drain resistor. Using the rubber eraser end of a pencil, I apply pressure to the SOT23 part for good electrical contact. The hard part is waiting for it to thermally stabilize before taking the reading. In a lot of 40 I can usually find 2 pairs that match currents better than 0.5% and 3 to 5 pairs that are within 1%. Sometimes one pair is within 0.2% - those are special and I set aside for my special top amps.
This thread has some good info:
Jfet matching jig
Thank you, I am trying to replace 170's with 862's, so this will become very useful. Meanwhile what is this pocket class A amp you talking about...can you kindly share some details or suggest link/forum
Thread:
xrk971 Pocket Class A Headamp GB
PCBs or full amps available in my signature website.
Head-fi:
New Portable Amp - "Pocket Class A" by xrk971 - now available as complete PCB | Head-Fi.org
Headfonics review:
XRK Audio – Headfonics
Last edited:
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.