JPV said:
JPV,
I'll see what I can do to get the resolution better at 500 uA. It may happen automatically since I have two fewer 3xxx devices... If all else fails, I will collect static data points for VF vs IF at 100 uA and 500 uA.
John
Here's the Vbe tempco data for the 4281/4302:
NPN NJL4281 (Average of 5 units)
T Ib Vbe Tempco
------------------------------------------------------
27ºC 100uA 0.469V
27ºC 1mA 0.548V
27ºC 10mA 0.635V
27ºC 100mA 0.730V
100ºC 100uA 0.279V -2.60mV/ºC
100ºC 1mA 0.380V -2.30mV/ºC
100ºC 10mA 0.489V -2.00mV/ºC
100ºC 100mA 0.603V -1.74mV/ºC
PNP NJL4302 (Average of 5 units):
T Ib Vbe Tempco
------------------------------------------------------
27ºC 100uA 0.443V
27ºC 1mA 0.526V
27ºC 10mA 0.625V
27ºC 100mA 0.728V
100ºC 100uA 0.259V -2.52mV/ºC
100ºC 1mA 0.359V -2.29mV/ºC
100ºC 10mA 0.473V -2.08mV/ºC
100ºC 100mA 0.600V -1.75mV/ºC
So, with a standing bias of 100mA and an Hfe of between 80 and 250 (characteristic curves I've measured confirm this variation between units), we're looking at a base current of between 1.3mA and 400 uA, so idle tempco is roughly -2.15 to -2.50 mV/ºC for Vbe vs about 1.8 mV/ºC for diode Vf at 10 mA.
The biasing circuit I'll using has a pot in series with the diode string to set bias current, and a pot in parallel with a diode or two to set tempco... And a day spent chasing bias hot/cold/hot/cold until I get tempco and bias adjusted right.
Cheers,
John
NPN NJL4281 (Average of 5 units)
T Ib Vbe Tempco
------------------------------------------------------
27ºC 100uA 0.469V
27ºC 1mA 0.548V
27ºC 10mA 0.635V
27ºC 100mA 0.730V
100ºC 100uA 0.279V -2.60mV/ºC
100ºC 1mA 0.380V -2.30mV/ºC
100ºC 10mA 0.489V -2.00mV/ºC
100ºC 100mA 0.603V -1.74mV/ºC
PNP NJL4302 (Average of 5 units):
T Ib Vbe Tempco
------------------------------------------------------
27ºC 100uA 0.443V
27ºC 1mA 0.526V
27ºC 10mA 0.625V
27ºC 100mA 0.728V
100ºC 100uA 0.259V -2.52mV/ºC
100ºC 1mA 0.359V -2.29mV/ºC
100ºC 10mA 0.473V -2.08mV/ºC
100ºC 100mA 0.600V -1.75mV/ºC
So, with a standing bias of 100mA and an Hfe of between 80 and 250 (characteristic curves I've measured confirm this variation between units), we're looking at a base current of between 1.3mA and 400 uA, so idle tempco is roughly -2.15 to -2.50 mV/ºC for Vbe vs about 1.8 mV/ºC for diode Vf at 10 mA.
The biasing circuit I'll using has a pot in series with the diode string to set bias current, and a pot in parallel with a diode or two to set tempco... And a day spent chasing bias hot/cold/hot/cold until I get tempco and bias adjusted right.
Cheers,
John
jgedde said:
Great!
Good point is that the temco of the Vbe of NPN and PNP are the same.
The temco of the Vbe rises ( in absolute value) by 0.2 to 0.3 mV/°C per decade of current increase which is correct.
The puzzeling point is that the temco of the Vbe of the diodes is between 1.7 and 1.8 mV/°C wathever the current. This shows a large departure from ideal.
Perhaps under 1mA will it be better but as such we cannot use the tuning of current to adjust temco of diodes. It is puzzeling
Taking carrefully the data from the Onsemi spec ( with a caliper) I have :
@1mA temco = -2.47mV/°C
@10mA -2.13
@100mA -1.95
This shows a normal change with current. I used the 100°C and the 25°C curves as I did with your data.
For the 3XXX, if you have a time problem, the interesting measurements would be some single points:
For 0.7mA Vbe at 25°C and 100°C
same for 0.8ma 1ma and 5mA
So four measurements at 25°C and 4 at 100 °C
Thanks
JPV
DouglasSelf said:
Hi Doug,
Actually, I just used a DVM across the emitters.
However, you are correct about the limitation - the approach does not yield useful data when signal is going through the amplifier. So one must interrupt the signal to take the measurement. My main interest has always been in seeing what happens to the bias after a "hot" interval has occurred and then gone away.
I believe that the thermal variations in many output stages as a result of this kind of thermal shock can be responsible for degraded amplifier sound that may not show up in conventional lab testing where the amplifier may be properly biased most of the time.
Only a 10 degree C mistracking can cause the nominal 26 mV bias to be off by 22 mV. This can be especially disastrous if it is in the direction of under-biasing the output stage. Although over-bias and consequent gm-doubling is to be avoided, it is, in my experience, less audibly bad than under-bias.
Cheers,
Bob
If I look at the Temco zip file I find that the diode tempco for 10mA is 1,89mV.
I have only looked at 10mA, but will look at other currents as well.
Stinius
I have only looked at 10mA, but will look at other currents as well.
Stinius
Sorry I made a mistake.
The data you gave are for base currents and not for collector currents. It is more difficult to interprete because of Hfe variation with temperature.
Do you have Vbe data at fixed 100mA and 10 mA collector current for 25°C and 100°C temperatures. These are 4 static measurements. The 10 mA is to make the Pease check
Thanks
JPV
The data you gave are for base currents and not for collector currents. It is more difficult to interprete because of Hfe variation with temperature.
Do you have Vbe data at fixed 100mA and 10 mA collector current for 25°C and 100°C temperatures. These are 4 static measurements. The 10 mA is to make the Pease check
Thanks
JPV
stinius said:
The tempco (-2mV/°C) is defined at Ic ( collector) constant. It is also in this mode that it is used with a Vbe spreader/stabilizer.
Stepping with Ib gives difficult to use data because hfe is changing with T in an unpredictable ( analytically) way within and from sample to sample.
JPV
JPV said:
Yes I know that, but it seems like John is measuring the Vbe with constant Ib and not constant Ic. It would be best if he used constant Ic.
Stinius
I have put the raw data from John in to an Excel spread sheet so it will be easier to find the different values, but it seem that I’m not able to attach the file, so if somebody wants the file I can e-mail it.
Stinius
Stinius
All,
I have started temperature testing of JPV's 3xxx series ThermalTrak diodes. -25 is complete. What I can tell you at this point is that it appears the diodes in 3xxx are definitely not the same as those in the 4xxx series...
The -25 curve agrees very well with the curve in the NJL3281/1302 datasheet. I was succesful in modifying the test setup to collect data down to 100 uA...
John
I have started temperature testing of JPV's 3xxx series ThermalTrak diodes. -25 is complete. What I can tell you at this point is that it appears the diodes in 3xxx are definitely not the same as those in the 4xxx series...
The -25 curve agrees very well with the curve in the NJL3281/1302 datasheet. I was succesful in modifying the test setup to collect data down to 100 uA...
John
jgedde said:
Great!!
With you test tools can you measure Vbe drift under constant collector current ?
JPV
jgedde said:
I have ziped now, but the size is 151Kb and it seems the max size is 100Kb.
So I will have to email them.
Just send me an e-mail and I will send the file.
Stinius
- Status
- Not open for further replies.