Hi.
I have two questions pertaining to Transistor matching.
1). Should transistors only with same batch numbers be matched and used together.
2). Why do some people match @/forthe idle current of the amplifier.
Thanx in advance all.
I have two questions pertaining to Transistor matching.
1). Should transistors only with same batch numbers be matched and used together.
2). Why do some people match @/forthe idle current of the amplifier.
Thanx in advance all.
You can use any devices with the same specification, i.e. label on the front.
If you test a To92 device at 1uA of base current and measure 100uA of collector current and then find a second device that also matches that hFE at that Ic then what happens if the Vbe of those two devices were different by say 10mVbe?
What happens if the two devices when operating in their circuit pass 3mA of Ic?
Does any of the testing at 1uA of Ib resulting in different Vbe give any guide to what happens in the circuit?
If you test a To92 device at 1uA of base current and measure 100uA of collector current and then find a second device that also matches that hFE at that Ic then what happens if the Vbe of those two devices were different by say 10mVbe?
What happens if the two devices when operating in their circuit pass 3mA of Ic?
Does any of the testing at 1uA of Ib resulting in different Vbe give any guide to what happens in the circuit?
Hi Andrew!
I don't understand what you're getting at. A little help still needed as to why people match at idle current.....
I don't understand what you're getting at. A little help still needed as to why people match at idle current.....
The matter is what you are matching and what for.
Small signal devices are often matched for both Vbe and Hfe (base-emitter diode voltage and DC gain of the device).
This is done because both qualities can affect the current balance in say, an LTP, where we want to minimise DC and AC
differences for optimum performance. Output transistors are matched for current sharing, which is quite hard to ensure
in big, multi-device amps. The idle current is a convenient, built-in reference temperature/current point relevant to the
amplifier that the devices will be used in, so when assembled, they should initially be sharing current as well as selection
process allows. This matching is also said to lower distortion significantly.
AFAIK, the same wafer/tube/reel sorting is convenient and often a quite adequate way of matching power devices without
the need to test each one. The testing can be quite tedious and requires a well designed jig, heatsinking and PSUs to have
reliable results. Commercially, it makes servicing more expensive so using manufactuiring consistency statistics can be
a great alternative. It relies on the consistency of die within a wafer/batch/ tube lot being within say, a few% for
Hfe & Vbe. A final check then confirms the adequacy of the matches.
Actually, with most small to medium DIY projects, only perfectionism drives a need to match BJTs. 😉
Small signal devices are often matched for both Vbe and Hfe (base-emitter diode voltage and DC gain of the device).
This is done because both qualities can affect the current balance in say, an LTP, where we want to minimise DC and AC
differences for optimum performance. Output transistors are matched for current sharing, which is quite hard to ensure
in big, multi-device amps. The idle current is a convenient, built-in reference temperature/current point relevant to the
amplifier that the devices will be used in, so when assembled, they should initially be sharing current as well as selection
process allows. This matching is also said to lower distortion significantly.
AFAIK, the same wafer/tube/reel sorting is convenient and often a quite adequate way of matching power devices without
the need to test each one. The testing can be quite tedious and requires a well designed jig, heatsinking and PSUs to have
reliable results. Commercially, it makes servicing more expensive so using manufactuiring consistency statistics can be
a great alternative. It relies on the consistency of die within a wafer/batch/ tube lot being within say, a few% for
Hfe & Vbe. A final check then confirms the adequacy of the matches.
Actually, with most small to medium DIY projects, only perfectionism drives a need to match BJTs. 😉
Last edited:
Thanx Ian. I have a test jig setup. Its basically Rod Eliott's jig, as is on his website. I try to match at the amplifier's quiscent current, but never knew why.
I only use it as an alternative to replacing all the OP devices, when doing repairs.
Matching at such low currents, doesnt require large hetsinks and fans.
Hugh R Dean, assured me that matching at between 50mA to 130ma of Ic, should be sufficient.
What's your take on this? Would like to hear more .....
I only use it as an alternative to replacing all the OP devices, when doing repairs.
Matching at such low currents, doesnt require large hetsinks and fans.
Hugh R Dean, assured me that matching at between 50mA to 130ma of Ic, should be sufficient.
What's your take on this? Would like to hear more .....
Hi
I haven't yet discussed matching with Hugh but I agree with his recommended range of matching bias current
The transistors have to be in their linear region to make useful comparisons and selection, so this puts them just
in range. When tested in-circuit, voltage and current may vary because of the number and type of junctions that
are engaged in the particular EF, CFP, or Quasi configuration. e.g. for an EF2 stage with high gain devices, I'd
like 100-120 mA for matching. It may require a fair heatsink to ensure the temp.stays low and cools quickly between
measurements for any accuracy, though.
D.Self explains bias dependencies better, in understandable terms in his design handbook. Still, I hope I'm making some sense here.
I haven't yet discussed matching with Hugh but I agree with his recommended range of matching bias current
The transistors have to be in their linear region to make useful comparisons and selection, so this puts them just
in range. When tested in-circuit, voltage and current may vary because of the number and type of junctions that
are engaged in the particular EF, CFP, or Quasi configuration. e.g. for an EF2 stage with high gain devices, I'd
like 100-120 mA for matching. It may require a fair heatsink to ensure the temp.stays low and cools quickly between
measurements for any accuracy, though.
D.Self explains bias dependencies better, in understandable terms in his design handbook. Still, I hope I'm making some sense here.
since most output stages are emitter followers, it seems to me that by matching for bias point, you can set up the situation where the devices track in terms of current delivered to the load - the danger is really only when the amp is driven hard, and in the case where one transistor is "on" way before the others or else has a rather different curve that it will "hog" the current, causing a failure.
Of course in theory if all devices were ideal and identical the circuit would operate in a more theoretical way, rather than the resulting Ib/Ic curve being a composite of all the devices working together...
I could be all wet on this, but that's how I have come to look at it.
_-_-bear
Of course in theory if all devices were ideal and identical the circuit would operate in a more theoretical way, rather than the resulting Ib/Ic curve being a composite of all the devices working together...
I could be all wet on this, but that's how I have come to look at it.
_-_-bear
just one little note -- Scott had mentioned -- when matching small signal devices try not to handle them too much as you'll transmit enough heat to the guys to cause the measurements to change.
- Status
- Not open for further replies.