Hello,
I'm almost certain that you will burn them, Are they in the VAS of an amp the might "see" almost twice the railvoltage. In this case it will be 120V.
They might not burn though. I used 139/140 in by Leach at +-55V rails before I changed to MJE340/350, the transistors played, but distorted like mad because of overvoltage.
\Jens
I'm almost certain that you will burn them, Are they in the VAS of an amp the might "see" almost twice the railvoltage. In this case it will be 120V.
They might not burn though. I used 139/140 in by Leach at +-55V rails before I changed to MJE340/350, the transistors played, but distorted like mad because of overvoltage.
\Jens
Even if they are used in a way that they should never see the rail-to-rail potential but only-to-common, some people routinely try to choose ratings that are twice what is "needed" just to allow for unforeseen circumstances. Some call this a "belt and suspenders" approach.
You could try 2SD669/2SB649 or 2SC2690/2SA1220A (KSC1690/KSC1220A, if Fairchild). If you can find them. . .
You could try 2SD669/2SB649 or 2SC2690/2SA1220A (KSC1690/KSC1220A, if Fairchild). If you can find them. . .
I've repaired amplifiers with BD139 & BD140 used in VAS and as pre-drivers with up to +-50V rails with no problems. I had to replace dozens of output devices but never had to replace a blown BD139 or BD140. At low currents [ie: 10mA] they appear to be able to reliably withstand 100V [te ones I saw were from Philips]
Anyway, for VAS or predriver with +-60V supplies I would look for 150V or higher Vceo transistors. BD139 & BD140 would be OK for other applications where they are exposed to a single 60V rail
JensRasmusen :
BD139 & BD140 are much faster than MJE340 & MJE350. That probably caused your leach amp to be unstable and oscillate thus producing very audible distortion
Anyway, for VAS or predriver with +-60V supplies I would look for 150V or higher Vceo transistors. BD139 & BD140 would be OK for other applications where they are exposed to a single 60V rail
JensRasmusen :
BD139 & BD140 are much faster than MJE340 & MJE350. That probably caused your leach amp to be unstable and oscillate thus producing very audible distortion
This is not the result of a carefully controlled test but I have gotten the impression the MJE340/350 may be a contributing factor toward oscillation in a CFB amp using 2sc5200/2sc1943 or similar. 2SD669/2SB649 "cured" the problem. While there is no reason to believe that this is directly applicable to the Leach, I mention only to support the observation posted, above, under some circumstances MJE340/350 may not be a good drop-in substitute.
In the above case a simiulation (not the most sophisticated in construction, I'm sure) showed a lot of high order harmonic spikes when the MJEs where in but not with others having higher fT, so there may be something to this.
In the above case a simiulation (not the most sophisticated in construction, I'm sure) showed a lot of high order harmonic spikes when the MJEs where in but not with others having higher fT, so there may be something to this.
there us no audible distorsion but i am still going to replace them when the mje340/50 are going to come .... just to be sure . 
sam9 :
Slow transistors in a VAS produce almost the same effect as increasing the value of C-B miller compensation capacitors. When the dominant pole of the amplifier is located on the VAS, lowering that pole reduces open-loop gain-bandwith product and increases phase margin thus improving stability [and increasing closed-loop THD due to less feedback]
In the other hand, slow transistors in a CFP contribute to oscillation, not to say that most CFP transistor combinations are inherently unstable when driven simulstaneously at high Ic and low Vce [ie: when 'big device' starts to saturate, at clipping threshold]
Jens :
Did you check wether this THD increase was due to transient RF oscillation [usually at zero-current crossing or output-device Ic/Vce dependent] due to too low phase margin, or due to C-B reverse breakdown leakage? [only above a certain output level threshold]. In case of reverse breakdown, the amplifier should show pretty low THD up to at least -6dB
Slow transistors in a VAS produce almost the same effect as increasing the value of C-B miller compensation capacitors. When the dominant pole of the amplifier is located on the VAS, lowering that pole reduces open-loop gain-bandwith product and increases phase margin thus improving stability [and increasing closed-loop THD due to less feedback]
In the other hand, slow transistors in a CFP contribute to oscillation, not to say that most CFP transistor combinations are inherently unstable when driven simulstaneously at high Ic and low Vce [ie: when 'big device' starts to saturate, at clipping threshold]
Jens :
Did you check wether this THD increase was due to transient RF oscillation [usually at zero-current crossing or output-device Ic/Vce dependent] due to too low phase margin, or due to C-B reverse breakdown leakage? [only above a certain output level threshold]. In case of reverse breakdown, the amplifier should show pretty low THD up to at least -6dB
Eva,
Although I'm not unfamiliar with the interplay between THD and OLG and dominate pole, I had not really appreciated the role of fast vs. slow transistors in that location. Your post states it succinctly and is quite helpful.
Re CFB: I have a "love-hate" relationship with them. I've built two nice amps with CFB outputs but I've struggled with another to the point of dumping the idea and using an EF layout. Apart from THD issues CFB has some nice attractions regarding construction - thermal tracking can be done more elegantly, the output devices don't need to all be on the same heatsink and they run a little cooler at no or low input.
Although I'm not unfamiliar with the interplay between THD and OLG and dominate pole, I had not really appreciated the role of fast vs. slow transistors in that location. Your post states it succinctly and is quite helpful.
Re CFB: I have a "love-hate" relationship with them. I've built two nice amps with CFB outputs but I've struggled with another to the point of dumping the idea and using an EF layout. Apart from THD issues CFB has some nice attractions regarding construction - thermal tracking can be done more elegantly, the output devices don't need to all be on the same heatsink and they run a little cooler at no or low input.
B.I.G said:can i go overvoltage for some BD139/140 ?
I have a +/-60V supply and they are rated 80Vceo ... I have used BD237/8 on +/-60V rails and had no problems and they are also rated 80Vceo
If you have access to a curve tracer, you can test what the actual Vceo of each device is. Some types and or batches will support higher voltage, some won't. It can vary from manufacturer and date code.
I've found many Japanese transistors to have much higher voltage capability than rated, but there were a few devices in each batch that were close to the spec limit too. I sometimes test and label a whole batch just as an experiment, and reserve the better devices for higher voltage application.
Bottom line, unless you test it's Russian Roulette, and with expensive output transistors as your bet. You have to decide if it's worth the risks.
Eva
I’m certain, that I did not have any oscillation of any kind. I checked with a scope and found no evidence of bad behaviour related to oscillation.
I did not investigate further, with all else being the same I changed the transistors from the BD pair to the MJE pair, and the distortion at 1 kHz and 0 dB went down a lot. At the time I had no time to measure further, and the amp has played since then (almost a year now)
What are the alternatives to the MJE pair? Same sort of voltage rating and power? It’s for VAS stages and pre driver in a three deep darlington configuration
\Jens
I’m certain, that I did not have any oscillation of any kind. I checked with a scope and found no evidence of bad behaviour related to oscillation.
I did not investigate further, with all else being the same I changed the transistors from the BD pair to the MJE pair, and the distortion at 1 kHz and 0 dB went down a lot. At the time I had no time to measure further, and the amp has played since then (almost a year now)
What are the alternatives to the MJE pair? Same sort of voltage rating and power? It’s for VAS stages and pre driver in a three deep darlington configuration
\Jens
I like Toshiba stuff....
http://www.diyaudio.com/forums/showthread.php?postid=390369#post390369
I like also BC550C/BC560C [small signal 45V], BC546B/BC556B [small signal 65V], 2N5401/2N5551 [small signal 150V, use several paralelled if required] and BF871/BF872 [small signal 300V]
Some time ago, I built an amplifier prototype using *only* BC550C/BC560C, 2SA1837/2SC4739 and 2SA1295/2SC3264 [Sanken 230V 17A and FT about 50Mhz]. I tested it with +-24V, +-56V and +-85V floating output rails [floating ground topology, QSC-like but modified] and used fixed +-15V for the front-end. In simulation it showed about 90dB open loop gain at 10Khz and the real prototype was still stable with 30dB closed loop gain [I used tricky frequency compensation with pole-zero shelvings to roll off the gain faster at about 9dB/oct just above audio the band]. That thing had enough gain-bandwith product to see 7 to 14Mhz oscillation on the output until I optimized a bit the layout
I think this demonstrates what can be achieved with modern fast high-power bipolars [since fast small-signal bipolars are pretty old stuff]
But look at Leach Amp plans : C10 and C11 are only 10pF and Q12-Q13 actually may have higher internal C-B capacitances. For example, 2N5415 datasheet states 25pF C-B capacitance at Vc-b=10V and I've just measured 50pF in a Philips MJE340 and 90pF in a Motorola MJE350 also with Vc-b=10V. Replacing VAS transistors is tricky since it's required to adjust also miller compensation capacitors depending on the difference of C-B capacitance between the old and the new device. With MJE340/MJE350 as VAS you may remove miller capacitors with no stability loss, the C-B capacitance of these transistors is much higher and dominates. In the other hand, with lower capacitance/faster transistors you may need to increase capacitor value
http://www.diyaudio.com/forums/showthread.php?postid=390369#post390369
I like also BC550C/BC560C [small signal 45V], BC546B/BC556B [small signal 65V], 2N5401/2N5551 [small signal 150V, use several paralelled if required] and BF871/BF872 [small signal 300V]
Some time ago, I built an amplifier prototype using *only* BC550C/BC560C, 2SA1837/2SC4739 and 2SA1295/2SC3264 [Sanken 230V 17A and FT about 50Mhz]. I tested it with +-24V, +-56V and +-85V floating output rails [floating ground topology, QSC-like but modified] and used fixed +-15V for the front-end. In simulation it showed about 90dB open loop gain at 10Khz and the real prototype was still stable with 30dB closed loop gain [I used tricky frequency compensation with pole-zero shelvings to roll off the gain faster at about 9dB/oct just above audio the band]. That thing had enough gain-bandwith product to see 7 to 14Mhz oscillation on the output until I optimized a bit the layout
I think this demonstrates what can be achieved with modern fast high-power bipolars [since fast small-signal bipolars are pretty old stuff]
But look at Leach Amp plans : C10 and C11 are only 10pF and Q12-Q13 actually may have higher internal C-B capacitances. For example, 2N5415 datasheet states 25pF C-B capacitance at Vc-b=10V and I've just measured 50pF in a Philips MJE340 and 90pF in a Motorola MJE350 also with Vc-b=10V. Replacing VAS transistors is tricky since it's required to adjust also miller compensation capacitors depending on the difference of C-B capacitance between the old and the new device. With MJE340/MJE350 as VAS you may remove miller capacitors with no stability loss, the C-B capacitance of these transistors is much higher and dominates. In the other hand, with lower capacitance/faster transistors you may need to increase capacitor value
Stability has, as you pointed out, a lot to do with the open loop gain of the amp. The (modified) Leach amp that I use has an open loop gain of about 65 dB and the dominating pole at about 5 kHz. The phase margin is about 100 degrees (simulated with the input filter bypassed) so there should be no problem with stability. The reason I chose to have such low OL gain is simply to try if is sounds nice – and I think it does.
I made some minor changes to component values to arrive at the OL gain that I did (Trial and error in the simulator)
Does anybody have a nifty way of measuring the OL response in real life – I thought about shorting the NFB signal to ground trough a big cap, but dismissed the idea because I’m afraid of the consequences should the amp go nuts
\Jens
I made some minor changes to component values to arrive at the OL gain that I did (Trial and error in the simulator)
Does anybody have a nifty way of measuring the OL response in real life – I thought about shorting the NFB signal to ground trough a big cap, but dismissed the idea because I’m afraid of the consequences should the amp go nuts
\Jens
Hi Jens,
I think your idea is not so bad.
You could split the feedback resistor into two resistors of the same
value and put a big cap from the center tap to ground.
DC-feedback loop will remain fine by this. AC feedback signal
is bypassed to ground.
Probably you should also divide your input signal, because you will need very small input signals.
Bye
Markus
@Destroyer X:
I know what you are thinking now
Probably you are right
I think your idea is not so bad.
You could split the feedback resistor into two resistors of the same
value and put a big cap from the center tap to ground.
DC-feedback loop will remain fine by this. AC feedback signal
is bypassed to ground.
Probably you should also divide your input signal, because you will need very small input signals.
Bye
Markus
@Destroyer X:
I know what you are thinking now
Probably you are right
BD139 are hardly used in my bench, to substitute hard to find Radio frequency drivers
They use to work with 500 miliamps, producing 16 volts output at 29 Megahertz... it is also used in FM link transmitters, going 122 Megahertz, producing 2 watts of radio frequency over 50 ohms
In the Wave form monitor, wave do not show nothing special, very clear signal...and watching spectrum analiser, it shows the harmonics that we normally find in Radio Frequency class A transmitters, those harmonics are normally filtered using double "T" shaped tuned filters...producing Greek letter that sound as "P" (3,14).
Those applications are 13.8 volts only, that's the main difference, that never shown me nothing wrong.
Very interesting Jens research, as working with high voltages related it's limits can strongly distort, compared to MJE340/350
Good that experience for us...thanks Jens.
- Chocoholic, i am also feeling alike you, so... we turn partners in that sittuation.
regards,
Carlos
They use to work with 500 miliamps, producing 16 volts output at 29 Megahertz... it is also used in FM link transmitters, going 122 Megahertz, producing 2 watts of radio frequency over 50 ohms
In the Wave form monitor, wave do not show nothing special, very clear signal...and watching spectrum analiser, it shows the harmonics that we normally find in Radio Frequency class A transmitters, those harmonics are normally filtered using double "T" shaped tuned filters...producing Greek letter that sound as "P" (3,14).
Those applications are 13.8 volts only, that's the main difference, that never shown me nothing wrong.
Very interesting Jens research, as working with high voltages related it's limits can strongly distort, compared to MJE340/350
Good that experience for us...thanks Jens.
- Chocoholic, i am also feeling alike you, so... we turn partners in that sittuation.
regards,
Carlos
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.