Hello
I`m buildin amplifier with bipolar onsemi transistors.
http://sound.westhost.com/project68.htm
I`v measured Hfe values with multimeter and all NPN transistors are 120+-1 and PNP 160+-1.
Now i don`t have a clue that should all transistors be same Hfe value or does it matter if PNP and NPN have different gain valua?
regards jyri
I`m buildin amplifier with bipolar onsemi transistors.
http://sound.westhost.com/project68.htm
I`v measured Hfe values with multimeter and all NPN transistors are 120+-1 and PNP 160+-1.
Now i don`t have a clue that should all transistors be same Hfe value or does it matter if PNP and NPN have different gain valua?
regards jyri
Ideally, the NPN and PNP parts would be precisely matched, howerver in practice it is futile to try, since a lot more than Hfe differs between them.
It would definitely be worth matching Q2 and Q3, but none of the others matter much.
It would definitely be worth matching Q2 and Q3, but none of the others matter much.
I assume from those gain values that you are talking about small signal transistors. For P68 the NPN/PNP gains do not have to be that close as the design is not symmetrical. It's good to have close values for each type though, and it looks like you have nicely matched devices 🙂
Only the output stage (which is symmetrical) should have matched gains, in an ideal world. But there are other factors that are more important here such as transfer characteristic.
Only the output stage (which is symmetrical) should have matched gains, in an ideal world. But there are other factors that are more important here such as transfer characteristic.
I`m building dual versio of the amplifier.
Transistors which i measured are Q9,Q11,Q9,Q11 Hfe 120+-1 (NPN)
and Q10,Q12,Q10,Q12 Hfe 160+-1 (PNP).
I have lot of these transistors.
What kind of gain should i search for Q7 and Q8?
Transistors which i measured are Q9,Q11,Q9,Q11 Hfe 120+-1 (NPN)
and Q10,Q12,Q10,Q12 Hfe 160+-1 (PNP).
I have lot of these transistors.
What kind of gain should i search for Q7 and Q8?
transfer characteristic.
richie could you explain what you mean by that.
Measured values are from output stage transistors.
richie could you explain what you mean by that.
Measured values are from output stage transistors.
k1jroth said:
Yipe.
Either they've already been matched or your meter is dead. In any case, you've certainly got the temperature even on those...
Tim
A lot of multimeters have a set of sockets where you can plug in small signal transistors and measure their HFE. The rub is knowing what collector current is used for this measurement, as HFE can vary quite a bit over several decades of collector current. For output transistors, I would think that VBE at the intended quiescent current would be a more important measurement, as it will allow one to more accurately predict the needed values for a bias network. As for other transistors, VBE matching is important for an input differential pair, as this will ultimately determine the output offset one must deal with to reduce DC through the speakers. Output transistor HFE is important to know at the maximum intended peak output current for the design, as it will help one to size the driver transistors.
Transfer characteristic is the correlation between base current or voltage, and the collector current. If you look at the datasheet for your devices there will be a graph that explains it visually and it should be quite easy to understand.
A typical DVM will measure transistor HFE at a collector current of around 100ua. A typical bipolar output stage will run at a collector current of at least 50-100ma to help smooth out crossover distortion. A typical power transistor will have relatively low HFE at the low collector current, will peak at a few hundred milliamps to an amp or so , then start to sag at higher currents. Where the peaking actually takes place, and where the falloff actually takes place will depend partly on die construction and partly on die size. A larger die rated for higher current will genrally sustain a reasonable HFE value at higher currents than a smaller die rated for lower currents. As was stated before, a lot of information can be gleaned from studying the HFE vs current curves for a given device. The more complete data sheets will have these curves, or at the very least will give HFE values for several conditions of collector current and VCE.
I haven't used bipolar transistors in a differential input stage for a while (I prefer JFETs and sometimes MOSFETs), but a certain amount of emitter degeneration will help swamp out mismatches in device VBE. I wouldn't rule out matching, either, as bipolar transistor are relatively inexpensive, and and the VBE spread will be much tighter than the usual VGS spread for a given populaion of JFETs or MOSFETs. This makes it much easier to get a reasonably close matched pair. Any matching should be done if possible at the intended quiescent collector current for the device in question.
I try not to go overboard with the emitter resistors, as this also reduces down the gain of the input stage. There is a tradeoff involved - too little gain will increase the overall distortion due to reduced loop gain (the input stage itself will be more linear, however, due to the local feedback), but at the same time will make the input stage less prone to transient overload. Operating with no degeneration at all gives the maximum gain, but there is the possibility of transient overload sensitivity and oscillation, especially if the input stage is run with fairly high current to get the slew rate up. I had one amp about 20 years ago that liked to take off and oscillate at a couple of MHz or so. This had nothing to do with the global compensation, and nothing I changed in that respect would help. I finally traced it to the input differential stage, which was running somewhere between 0.5-1ma/side. Addiing 22 ohms of emitter degeneration to each transistor quieted the input stage down quite satisfactorily, and the amp did yeoman service in my car for quite a few years.
I haven't used bipolar transistors in a differential input stage for a while (I prefer JFETs and sometimes MOSFETs), but a certain amount of emitter degeneration will help swamp out mismatches in device VBE. I wouldn't rule out matching, either, as bipolar transistor are relatively inexpensive, and and the VBE spread will be much tighter than the usual VGS spread for a given populaion of JFETs or MOSFETs. This makes it much easier to get a reasonably close matched pair. Any matching should be done if possible at the intended quiescent collector current for the device in question.
I try not to go overboard with the emitter resistors, as this also reduces down the gain of the input stage. There is a tradeoff involved - too little gain will increase the overall distortion due to reduced loop gain (the input stage itself will be more linear, however, due to the local feedback), but at the same time will make the input stage less prone to transient overload. Operating with no degeneration at all gives the maximum gain, but there is the possibility of transient overload sensitivity and oscillation, especially if the input stage is run with fairly high current to get the slew rate up. I had one amp about 20 years ago that liked to take off and oscillate at a couple of MHz or so. This had nothing to do with the global compensation, and nothing I changed in that respect would help. I finally traced it to the input differential stage, which was running somewhere between 0.5-1ma/side. Addiing 22 ohms of emitter degeneration to each transistor quieted the input stage down quite satisfactorily, and the amp did yeoman service in my car for quite a few years.
Matching is pretty unnecessary and if you should match anything it should be Q2 and Q3 but I doubt that you will need it.k1jroth said:
One easy way of matching is to solder one transistor and then you solder in a slaughtered DIL socket with round holes. You can also get a socket row which is cuttable. Solder in those three socket contacts and then get 5-10-100! transistors. Put in one transistor, write down the offset you have got. Then you take as many transistors you have and test. When you have found a good value take this transistor and solder it in properly.
I just matched Vbe on a lot of 100 C2240, and the range was only 2mV. I feel like I may have wasted my time slightly.
You did pretty well with the C2240s probably because they were all from the same production lot, possibly even from the same wafer. A mixed lot would most likely show a wider spread. You did not really waste your time, as now you know for certain you can pull a couple of transistors at random from that batch and construct a well-matched differential pair.
If I get really fussy about matching and drift, I lay out the board so that I can place the pair in question in face-to face contact (flat face of TO-92 package). I put a little thermal grease on the package faces and then bind them together with a few turns of self-adhesive copper tape. This is being extra fussy, and I do not normally do this unless the matched pair is expected to dissipate significant power. This is usually not an issue with a bipolar front end, but it can be with a MOSFET front end where one is running a higher drain current to get better linearity. In this case, you want the differential pair to be in the same thermal environment to avoid drift. I recently did this trick with a pair of matched MPF960 TO-92L MOSFETs used as a differential front end.
If I get really fussy about matching and drift, I lay out the board so that I can place the pair in question in face-to face contact (flat face of TO-92 package). I put a little thermal grease on the package faces and then bind them together with a few turns of self-adhesive copper tape. This is being extra fussy, and I do not normally do this unless the matched pair is expected to dissipate significant power. This is usually not an issue with a bipolar front end, but it can be with a MOSFET front end where one is running a higher drain current to get better linearity. In this case, you want the differential pair to be in the same thermal environment to avoid drift. I recently did this trick with a pair of matched MPF960 TO-92L MOSFETs used as a differential front end.
Added schematic which i think is good way to measure tarnsistors
VBE or Hfe value.
explain of schematic:
if i measure current between R2 and base
and current between emitter to GND
and calculate Ie/Ib=Hfe
I think it should give me enough accurate value for Hfe.
Then i just change transistor and do not touch resistors or input voltage
I think i can found matched components with this measurement.
correct if i`m wrong.
Does it matter if i measure it like this or should i measure Vbe?
Why?
no one still answered my question about that should all output stage transistors match or is it good enough if i match all NPN (Q7,Q9,Q11 same value) and then all PNP (Q8,Q10,Q12 same value and different than NPN transistors.
I think that if all should match together then allso Q4 and Q6 should be matched (differential stage)
I´m kind of beginner in audio electronics so please explain simply.
VBE or Hfe value.
explain of schematic:
if i measure current between R2 and base
and current between emitter to GND
and calculate Ie/Ib=Hfe
I think it should give me enough accurate value for Hfe.
Then i just change transistor and do not touch resistors or input voltage
I think i can found matched components with this measurement.
correct if i`m wrong.
Does it matter if i measure it like this or should i measure Vbe?
Why?
no one still answered my question about that should all output stage transistors match or is it good enough if i match all NPN (Q7,Q9,Q11 same value) and then all PNP (Q8,Q10,Q12 same value and different than NPN transistors.
I think that if all should match together then allso Q4 and Q6 should be matched (differential stage)
I´m kind of beginner in audio electronics so please explain simply.
Attachments
Your circuit is OK but you will need to use much more than 0.67 ohms in the collector to avoid risk of destruction of both the resistor and transistor. You should also maybe only put 0.5-2 volts on the base. You could be running the device into saturation like you have it, which would give misleading results.
You should also calculate gain as Ic/Ib not Ie as the base current flows out of the emitter so you will get a misleading result.
You should also calculate gain as Ic/Ib not Ie as the base current flows out of the emitter so you will get a misleading result.
if i put 534ohm resistor series with base it lovers maximum current to 37,5mA base can take 1,5A and 5Vdc.
I don`t understand why it would broke transistor.
i made mistake calculating collector resistor i think
20ohm 20W wirewound should do it
It limits maximum current to 1A.
with 1A collector current saturation is below 0,8V so it`s pretty low.
Pleso correct if i`m wrong (again).
richie yep it should be Ic my mistake
I don`t understand why it would broke transistor.
i made mistake calculating collector resistor i think
20ohm 20W wirewound should do it
It limits maximum current to 1A.
with 1A collector current saturation is below 0,8V so it`s pretty low.
Pleso correct if i`m wrong (again).
richie yep it should be Ic my mistake
You should get rid of the collector resistor, and program a known current through the emitter. Typically this would be the quiescent current of the circuit in which you are using the transistor. 1mA, for example.
Use a large resistor for the base, say 10k. Then program a current through the base resistor, such as 100µA. The change in voltage on the base resistor will tell you the Hfe, and you can measure Vbe directly.
Use a large resistor for the base, say 10k. Then program a current through the base resistor, such as 100µA. The change in voltage on the base resistor will tell you the Hfe, and you can measure Vbe directly.
- Status
- Not open for further replies.