lindsay:
can you post a pic? jp isn't a suffix that describes any idss ratings (unless things have changed recently)
are you sure there is not a "GR" somewhere on the front?
🙂
mlloyd1
can you post a pic? jp isn't a suffix that describes any idss ratings (unless things have changed recently)
are you sure there is not a "GR" somewhere on the front?
🙂
mlloyd1
LHMAudio said:
RF sensitivity:
Some people actually use a bjt-input opamp as an AM receiver:
http://www.i-trixx.com/uk/index.php?do=ezine&ed=080625_13801
(It looks as if the LM386 is only used to amplify RF, but in the text they admit that it also does its part of the demodulation).
Jan Didden
Some people actually use a bjt-input opamp as an AM receiver:
http://www.i-trixx.com/uk/index.php?do=ezine&ed=080625_13801
(It looks as if the LM386 is only used to amplify RF, but in the text they admit that it also does its part of the demodulation).
Jan Didden
MCM JFETs
Lindsay -
The FETs I received from Newark (MCM) last week are all sufffix BL (6-12mA IDss) as printed on the devices themselves.
The antistat bags were simply labeled 2SJ74 JP (I assume for Japan as country of origin?)
LHMAudio said:
Lindsay -
The FETs I received from Newark (MCM) last week are all sufffix BL (6-12mA IDss) as printed on the devices themselves.
The antistat bags were simply labeled 2SJ74 JP (I assume for Japan as country of origin?)
RFI
Here are some RFI simulations on the fron-end of a couple different amps. Notice that the output stage is replaced by an ideal buffer. Also notice that these amps have different tail currents. As it appeared that a higher tails current results in less AM demodulation, one should take this into account.
The following amps have been tested:
1. Bob's EC-amps, two versions, one with BJTs at the input, the other with JFETs. I had had to introduce some minor modifications to suppress instabilities: C3 increased from 150 to 220pF, R14 decreased from 100 to 47 Ohm and the VAS loaded with 10pF and 100 Ohm in series. The JFET version, however, still gave some unrealistic THD figures. After replacing the NPD5564 by a 2SK389, the problems were gone. Tail current: 4mA
2. D. Self's blameless amp. Tail current: 6mA
3. The PGP amp. Tail current: 2mA
4. The PCP amp (improved version of PMP). Sum of IPS collector currents: 2.6mA (this amp has no tail, instead a CFB IP stage)
RFI signal: Fc=10MHz, A=0.1Vpk, Fm=20kHz and m=0.7, applied to the non-inverting input. Input filters disabled of course.
The following AM demodulation products (20kHz) were observed.
Bob-JFET:. 0.722mV
Bob-BJT:.. 0.311mV
Self:........ 0.170mV
PGP:........ 5.10mV (0.85mV @4mA tail current)
PCP:........ 1.1mV
The reader may draw his own conclusions from these figures, that is, as long as he don't think JFETs are always better. 😀
Next, I've also looked at the effect on THD20 (in ppm) of an un-modulated carrier of 10MHz and 0.1Vpk.
The AF signal is 1Vpk and 20kHz.
..............no carrier... with carrier
Bob-JFET:. 0.233 ....... 0.260
Bob-BJT:.. 0.409 ........ 0.445
Self:........ 2.9 ........... 3.2
PGP:........ 0.039 ....... 0.024
PCP:........ 0.018 ....... 0.031
As one can see, the THD figures are hardly compromised by the HF carrier, but remember, the amplitude is "only" 0.1V. At 1V we get a different picture. I've only tested the PCP amp equipped with true (although non-existing) complementary trannies.
0.1V carrier: AM product = 0.37nV. THD20 rose from 4.7ppb to 5.0ppb
1.0V carrier: AM product = 1.25uV. THD20 rose from 4.7ppb to 24ppm
AM demodulation is almost nihil. Not surprising, as the circuit is 100% symmetrical. But at strong RFI levels the THD rises considerably. That's precisely what Glen has predicted.
Cheers,
Edmond.
Here are some RFI simulations on the fron-end of a couple different amps. Notice that the output stage is replaced by an ideal buffer. Also notice that these amps have different tail currents. As it appeared that a higher tails current results in less AM demodulation, one should take this into account.
The following amps have been tested:
1. Bob's EC-amps, two versions, one with BJTs at the input, the other with JFETs. I had had to introduce some minor modifications to suppress instabilities: C3 increased from 150 to 220pF, R14 decreased from 100 to 47 Ohm and the VAS loaded with 10pF and 100 Ohm in series. The JFET version, however, still gave some unrealistic THD figures. After replacing the NPD5564 by a 2SK389, the problems were gone. Tail current: 4mA
2. D. Self's blameless amp. Tail current: 6mA
3. The PGP amp. Tail current: 2mA
4. The PCP amp (improved version of PMP). Sum of IPS collector currents: 2.6mA (this amp has no tail, instead a CFB IP stage)
RFI signal: Fc=10MHz, A=0.1Vpk, Fm=20kHz and m=0.7, applied to the non-inverting input. Input filters disabled of course.
The following AM demodulation products (20kHz) were observed.
Bob-JFET:. 0.722mV
Bob-BJT:.. 0.311mV
Self:........ 0.170mV
PGP:........ 5.10mV (0.85mV @4mA tail current)
PCP:........ 1.1mV
The reader may draw his own conclusions from these figures, that is, as long as he don't think JFETs are always better. 😀
Next, I've also looked at the effect on THD20 (in ppm) of an un-modulated carrier of 10MHz and 0.1Vpk.
The AF signal is 1Vpk and 20kHz.
..............no carrier... with carrier
Bob-JFET:. 0.233 ....... 0.260
Bob-BJT:.. 0.409 ........ 0.445
Self:........ 2.9 ........... 3.2
PGP:........ 0.039 ....... 0.024
PCP:........ 0.018 ....... 0.031
As one can see, the THD figures are hardly compromised by the HF carrier, but remember, the amplitude is "only" 0.1V. At 1V we get a different picture. I've only tested the PCP amp equipped with true (although non-existing) complementary trannies.
0.1V carrier: AM product = 0.37nV. THD20 rose from 4.7ppb to 5.0ppb
1.0V carrier: AM product = 1.25uV. THD20 rose from 4.7ppb to 24ppm
AM demodulation is almost nihil. Not surprising, as the circuit is 100% symmetrical. But at strong RFI levels the THD rises considerably. That's precisely what Glen has predicted.
Cheers,
Edmond.
I would like to follow up on the noise measurements of the BF862. I originally saw a number of 'measurements' that were all over the map, so to speak. The problem with this part, besides limited voltage breakdown, is that it is not SPEC'ED for low frequency low noise. This can be a big problem as one batch of devices might be very quiet, and the next batch could very well be very noisy, as there is no 'quality control' requirement imposed by the spec sheet on the device.
It is well known that 1/f noise is an almost uncontrollable variable in quality fets, and requires a good attention to fab detail to get it right. This is what the fabricators tell me, in any case.
Independent screening of a FEW samples of the BF862 has shown it to be a pretty fair device, with perhaps twice the 1/f noise of a typical 2SK170, BUT with lower input capacitance that would make it a specific contender for condenser microphone input electronics, among other uses, just like Scott Wurcer mentioned on this thread a few weeks ago.
This does not mean that EVERY batch will be as quiet, but since two independent batches have been independently sampled with Quanteks (an obsolete technology as stated here on this thread by others) it looks OK to try them, and include them in the few devices that we have left to work with.
It is well known that 1/f noise is an almost uncontrollable variable in quality fets, and requires a good attention to fab detail to get it right. This is what the fabricators tell me, in any case.
Independent screening of a FEW samples of the BF862 has shown it to be a pretty fair device, with perhaps twice the 1/f noise of a typical 2SK170, BUT with lower input capacitance that would make it a specific contender for condenser microphone input electronics, among other uses, just like Scott Wurcer mentioned on this thread a few weeks ago.
This does not mean that EVERY batch will be as quiet, but since two independent batches have been independently sampled with Quanteks (an obsolete technology as stated here on this thread by others) it looks OK to try them, and include them in the few devices that we have left to work with.
RFI
(continued from post #7124)
And here are the (simulated) results with a hefty RF input of 1Vpk.
AM demodulation:
Bob-JFET:. 1.5V
Bob-BJT:.. 2.15V
Self:........ 68mV
PGP:........ 4.44V (@2mA tail current)
PGP:........ 1.77V (@4mA tail current)
PCP:........ 2.74V
THD20:
Bob-JFET:. 148ppm
Bob-BJT:.. 395ppm
Self:........ 105ppm
PGP:........ 26.4ppm (@2mA tail current)
PGP:........ 5.07ppm (@4mA tail current)
PCP:........ 900ppm
As we can see, the distortion of the PGP amp is the least affected by this large RFI, probably because of the NDFL stage that attenuates RF before it enters the VAS.
The PCP amp, on the other hand, clearly gives up. Apparently due to mismatches between complementary devices, because with true complements (see previous post), the rise of THD is limited to 24ppm.
As for Bob's EC amp, these figures clearly demonstrate that JFETs perform better with regard to distortion as well with regard to AM demodulation.
NB: These results were obtained without any filtering at the input. So I think that in real life with good shielding etc., amplifier inputs will never subjected to the terror of a 1V 10MHz interference. Therefore, the figures presented in my previous post may be more representative.
Cheers,
Edmond.
(continued from post #7124)
And here are the (simulated) results with a hefty RF input of 1Vpk.
AM demodulation:
Bob-JFET:. 1.5V
Bob-BJT:.. 2.15V
Self:........ 68mV
PGP:........ 4.44V (@2mA tail current)
PGP:........ 1.77V (@4mA tail current)
PCP:........ 2.74V
THD20:
Bob-JFET:. 148ppm
Bob-BJT:.. 395ppm
Self:........ 105ppm
PGP:........ 26.4ppm (@2mA tail current)
PGP:........ 5.07ppm (@4mA tail current)
PCP:........ 900ppm
As we can see, the distortion of the PGP amp is the least affected by this large RFI, probably because of the NDFL stage that attenuates RF before it enters the VAS.
The PCP amp, on the other hand, clearly gives up. Apparently due to mismatches between complementary devices, because with true complements (see previous post), the rise of THD is limited to 24ppm.
As for Bob's EC amp, these figures clearly demonstrate that JFETs perform better with regard to distortion as well with regard to AM demodulation.
NB: These results were obtained without any filtering at the input. So I think that in real life with good shielding etc., amplifier inputs will never subjected to the terror of a 1V 10MHz interference. Therefore, the figures presented in my previous post may be more representative.
Cheers,
Edmond.
Speaking of RFI I had reason to tell someone to try the lopsided Tom Holman input stage. Anyone ever try a complimentary version of that? I also found a recent (if you can believe that) very favorable review of a refurbished APT preamp, with the usual "you would have to spend X1000's of dollars today to equal its performance".
Once I got the FETS under good light with a magnifying glass I could see the grade markings.
So what do these markings mean?
2SK170 BL
2SJ74 GR
Thanks,
Lindsay
So what do these markings mean?
2SK170 BL
2SJ74 GR
Thanks,
Lindsay
scott wurcer said:
Thanks Scott.
I had a quick look at it and wonders what's so special about it?
Cheers,
Edmond.
I wonder if ALL the p channel fets (2SJ74) from MCM in this final buy are GR? IF they are then we are in trouble.
RFI - errata
Hi,
Actually, the RFI figures of the PCP amp at 1V RFI aren't that bad. I simply forgot to disable the current limiters of the VAS stages. Without these limiters I go a quite different picture:
PCP amp with lousy complementary devices:
AM = 1.78V and THD20 = 31.3ppm
PCP amp with true complementary devices:
AM = 2.4uV and THD20 = 13ppb.
CFB-BJT input stages rule!
Cheers,
Edmond.
Hi,
Actually, the RFI figures of the PCP amp at 1V RFI aren't that bad. I simply forgot to disable the current limiters of the VAS stages. Without these limiters I go a quite different picture:
PCP amp with lousy complementary devices:
AM = 1.78V and THD20 = 31.3ppm
PCP amp with true complementary devices:
AM = 2.4uV and THD20 = 13ppb.
CFB-BJT input stages rule!
Cheers,
Edmond.
John, I was wondering about matching the different Idss, but did not want to ask a stupid question. Seems there would be very little overlap between the GR and BL. Should I return them? Thanks.
Lindsay
Lindsay
Don't return them, but you you might have to try to buy some 2SK170 GR to best match. In any case, you can use what you have. I will explain later, if necessary. Still, it might be that Toshiba is 'cherry picking' parts for Chas Hansen, and we may not be able to get 2SJ74BL through normal channels, easily.
john curl said:
I've asked, begged, moaned, bitched at MCM and that's it. All they have is 2SJ74GR and 2SK170BL. I have my stock of 2SJ74BL so I'm fortunately trouble free for the next ten years.
For what it's worth, I sorted a batch of GR's and BL's and the
bulk of the GR's were about 6 ma, and the bulk of the BL's were
8 mA. Not so bad.
bulk of the GR's were about 6 ma, and the bulk of the BL's were
8 mA. Not so bad.
- Status
- Not open for further replies.