Since no experts are stepping in, I'll give you an answer based on my understanding (warning: could be wrong,,, I take this as an opportunity for learning).
For JFET, the relationship of Idss and Vgs is a function of temperature. in another word, for same make of JFET, you can calculate Idss from Vgs at certain temperature. So, matching either with Idss or Vgs is fine (given that you can manage the temperature of the junction very well). If you measure the Vgs at the Ids for the zero temperature coefficient point , then you are free from the effect of temperature change. The TC=0 point is dependent on the make of JFET, no plot found on many data sheets, impractical for some models and
By the way, a question for experts: Is there a way to calculate TC0 point in SPICE??
For biopolars, the relationship between hfe and vbe is not fixed, so people pick one or other dependent on their circuit requirements. Of course, you can match them for both parameters...
Even how hard you match two devices, the loose temperature coupling makes them wander off from each other. You have to use monolithic matched pair transistors if you don't want that to happen. Analog Devices' MAT series are fine matched low noise pairs (but expensive...). For quad matched low noise PNP, THAT 320 is almost the only choice (again expensive). For matched pairs on two separate dies, Diodes Inc has nice and cheap matched (<2% for both Vbe and hfe) transistor pairs but their noise figure is not as good as MAT or THAT and also they are only available in SMT. Linear Integrated Systems offers various matched transistors in variety of packages but I don't know the availability.
What's the tolerance for mismatch for your circuit? How about temperature drift?
Have fun!
Satoru
Hi Satoru
Thank you for the information. They are very useful and helpful for me to match transistors. I do not have a specify schematic or circuit to build yet but I want to match transistors at input differential and output stages for my up coming pre & power amp projects.
Thank you for the information. They are very useful and helpful for me to match transistors. I do not have a specify schematic or circuit to build yet but I want to match transistors at input differential and output stages for my up coming pre & power amp projects.
SSDIY,
Have you built any discrete solid state amps before? I'm not sure your emphasis on matching transistors is all that critical to circuit performance or reliability. I think you'll find that the impact of proper board layout, component ratings and thermal calculations will far and away impact reliability and DC offset more than transistor matching. This is especially true for three stage Lin amps and amps that have been influanced by Self and Sloane among others.
Besides matching output devices at 100uA, as your PDF circuit will do, won't do you much good at all. You'd have to modify the circuit to draw more current...and you'd only be matched at one point anyway.
Scott
Hi SSDIY,
For input dual JFET for a pre-amp, jump on the Group Buy for LSK389B. It is a good, high gm, low noise dual JFET suited for pre-amp. It can be used for power amp, too, but I'd recommend uPA68H which you can find in swap meet (sold by StevenOH). It is a low gm, low drift dual JFET good for power amp input stage. If you want comple input stage, then get matched set of 2SK170/2SJ74 from Blues in swap meet. You can put them in a metal can which you still can buy from EUVL (am I right?), then you get 2SK240/2SJ75. Blues' matching is better than original 2SK240/2SJ75 from Toshiba so you'll be happy! You can build a power amp with 2SK170/2SJ74 using JC-type topology which I like a lot. It is recommended to have DC servo if you are going to use 2SK240 (with or without 2SJ75) for input stage. It can be passive or active, it's your choice. It is easier to build a high slew rate amp (>100 V/us) with JFET input, so start with some dual JFETs while they are still obtainable!
For output stage, matching is not that critical. Pick a pair of output device with good reputation and you'll be fine.
Have fun with building an amp!
Good luck!
Satoru
For input dual JFET for a pre-amp, jump on the Group Buy for LSK389B. It is a good, high gm, low noise dual JFET suited for pre-amp. It can be used for power amp, too, but I'd recommend uPA68H which you can find in swap meet (sold by StevenOH). It is a low gm, low drift dual JFET good for power amp input stage. If you want comple input stage, then get matched set of 2SK170/2SJ74 from Blues in swap meet. You can put them in a metal can which you still can buy from EUVL (am I right?), then you get 2SK240/2SJ75. Blues' matching is better than original 2SK240/2SJ75 from Toshiba so you'll be happy! You can build a power amp with 2SK170/2SJ74 using JC-type topology which I like a lot. It is recommended to have DC servo if you are going to use 2SK240 (with or without 2SJ75) for input stage. It can be passive or active, it's your choice. It is easier to build a high slew rate amp (>100 V/us) with JFET input, so start with some dual JFETs while they are still obtainable!
For output stage, matching is not that critical. Pick a pair of output device with good reputation and you'll be fine.
Have fun with building an amp!
Good luck!
Satoru
For bipolars:
Vbe is base emitter voltage, specified at a particular collector current and voltage between collector and emitter. On an audio amp, and seeking matches, it makes sense to match Vbe for the collector current normally chosen for output stage bias and collector emitter voltage, typically 50-150mA and 36-80V respectively. Then you are matching the Vbe for the precise same circuit conditions the devices will be operating in, which is important to temperature management and current share.
Hfe is the DC current gain (strongly related to the AC current gain, which is hfe). This is the ratio of collector current to base current, typically 50-150 for a large output device.
Since in an audio amp the output which lie in parallel are normally driven from the same driver, they receive the same drive, so it's important that they should all pass the same current as a result. This is very evident in quiescent conditions, when the amp output stage is idling at the bias current. While the emitter resistors, typically 0.22 ohm, force current sharing, it is also true that identical Hfe is even more important to ensure all outputs pass the same current.
So, for parallel output devices in audio amps, you really should:
1. Match Vbe for the bias current of the output stage
2. Further refine matches by ensuring same Hfe AND same Vbe
3. To keep dissipation unsinked under control, use the same LOW voltage across the CE of the device, say 6V, throughout testing.
4. Make all measurements swiftly to avoid heating up the device appreciably.'
If the devices heat up, Vbe drops, by approx 1.5mV/degree C. So use a DMM with a high sampling rate!
Hope this helps, won't comment on FETs, not my scene,
Hugh
Vbe is base emitter voltage, specified at a particular collector current and voltage between collector and emitter. On an audio amp, and seeking matches, it makes sense to match Vbe for the collector current normally chosen for output stage bias and collector emitter voltage, typically 50-150mA and 36-80V respectively. Then you are matching the Vbe for the precise same circuit conditions the devices will be operating in, which is important to temperature management and current share.
Hfe is the DC current gain (strongly related to the AC current gain, which is hfe). This is the ratio of collector current to base current, typically 50-150 for a large output device.
Since in an audio amp the output which lie in parallel are normally driven from the same driver, they receive the same drive, so it's important that they should all pass the same current as a result. This is very evident in quiescent conditions, when the amp output stage is idling at the bias current. While the emitter resistors, typically 0.22 ohm, force current sharing, it is also true that identical Hfe is even more important to ensure all outputs pass the same current.
So, for parallel output devices in audio amps, you really should:
1. Match Vbe for the bias current of the output stage
2. Further refine matches by ensuring same Hfe AND same Vbe
3. To keep dissipation unsinked under control, use the same LOW voltage across the CE of the device, say 6V, throughout testing.
4. Make all measurements swiftly to avoid heating up the device appreciably.'
If the devices heat up, Vbe drops, by approx 1.5mV/degree C. So use a DMM with a high sampling rate!
Hope this helps, won't comment on FETs, not my scene,
Hugh
Any luck of matching complementary outputs? I've got two pairs of 5200/1943, they mismatched greatly (something like 100:150). Is it possible to match PNP to NPN at all? Just don't want to see the same numbers with new ordered pairs...
Hi gentlemen
Thank you very much for all your inputs. Most of the information you posted here are something new and good knowing to me.
I used to measure the idss of JFets for differential input but the outcome was not good. It had 30~70mv offset at the output of the pre-amp I built. So this time I will use the method which peranders suggested on post #12.
I will keep up-dated once I have the result.
Thank you very much for all your inputs. Most of the information you posted here are something new and good knowing to me.
I used to measure the idss of JFets for differential input but the outcome was not good. It had 30~70mv offset at the output of the pre-amp I built. So this time I will use the method which peranders suggested on post #12.
I will keep up-dated once I have the result.
Hi sekhar
It is because very hard to find a super matched pair P Channel JFets, N Channel is not the problem.
- Status
- Not open for further replies.