Going to 2pair/3pair is already a redesign.MJL21193 said:
Changing the dual VAS and CCS types is an even bigger change to Mike's design.
I am not competant to advise.
I await reply from Mike re adding a second pair and what he thinks might be needed to compensate (I'll guarantee that some values need to be changed). I will be looking at your results to inform my experiment. I want to replace the amps inside an active Tannoy and drive an 8ohm treble and 4ohm bass/mid. (that's four PCBs accounted for already).
I your looking for more power look here. Post #399.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=87609&perpage=25&pagenumber=16
I'd check with Roender to make sure it's been fully tested.
http://www.diyaudio.com/forums/showthread.php?s=&threadid=87609&perpage=25&pagenumber=16
I'd check with Roender to make sure it's been fully tested.
AndrewT said:
Going with what I have on hand, using a BD140 for Q5 (my schematic) looks doable.
Hi,MJL21193 said:
it's not just Q5.
I think you need to look at Q4, 5, 8, 9 and maybe Q7.
Balance across the LTP is important to maintaining accuracy of operation.
That's my point, changing 4 or 5 transistors is a major redesign,
adding 4 transistors simply exacerbates this.
AAK said:
I'm trying to get the present design to work with my supply.
I'm not necessarily looking for more power, but to get great performance using the components I already have.
I only hurt myself if my redesign doesn't perform as simmed, but by just switching to the original components, I have original performance.
Besides, I like what I'm doing here. 🙂
AndrewT said:
Checking current flow through this stage in the redesign and comparing it to the original shows not a huge difference. The big difference in this area is voltage, up to 50% higher.
Using the higher voltage devices might be enough.
Hi Mjl,
I am unsure if you grasp the reactive load consequence.
check your voltages and currents in the quiescent mode.
when you multply the current times voltage across any active or passive device you will obtain the quiescent power dissipation.
Now look at maximum output signal.
Check the instantaneous voltage and current across the devices, again deriving the dissipation.
When the load is open circuit or resistive you generally find that as voltage increases current decreases and power dissipation moves around a bit.
However into a reactive load you usually find that as the voltage increases you can also have the current increasing as well and the result is a very large increase in instantaneous dissipation.
This is so at the outputs.
I usually use Vrail times half the peak current into my minimum load to get first guess on power to allow selection of devices.
Then do a more accurate SOAR analysis to fine tune the stage.
Next, I look at the drivers and apply a modified version of SOAR analysis. The drivers can be very highly stressed due to the output reflecting the reactive load back to the drivers.
I have never properly checked the pre-driver nor the VAS due to the huge safety factor that normally exists with these two components. I rely on that safety factor to leave sufficient dissipation capacity to absorb something of the order of three times the power during some parts of the signal cycle.
With a two stage EF the VAS sees a significant loading reflected through the outputs and drivers.
That is my worry with your proposal. low impedance and lower current gain will stress the VAS. Simply looking at maximum current and maximum voltage specifications gets you no where near the operating conditions.
I do not mean to be doom and gloom, but you must look deeper into what you propose. Can Spice help you identify the worst combination of instantaneous current and voltage in the VAS?
I am unsure if you grasp the reactive load consequence.
check your voltages and currents in the quiescent mode.
when you multply the current times voltage across any active or passive device you will obtain the quiescent power dissipation.
Now look at maximum output signal.
Check the instantaneous voltage and current across the devices, again deriving the dissipation.
When the load is open circuit or resistive you generally find that as voltage increases current decreases and power dissipation moves around a bit.
However into a reactive load you usually find that as the voltage increases you can also have the current increasing as well and the result is a very large increase in instantaneous dissipation.
This is so at the outputs.
I usually use Vrail times half the peak current into my minimum load to get first guess on power to allow selection of devices.
Then do a more accurate SOAR analysis to fine tune the stage.
Next, I look at the drivers and apply a modified version of SOAR analysis. The drivers can be very highly stressed due to the output reflecting the reactive load back to the drivers.
I have never properly checked the pre-driver nor the VAS due to the huge safety factor that normally exists with these two components. I rely on that safety factor to leave sufficient dissipation capacity to absorb something of the order of three times the power during some parts of the signal cycle.
With a two stage EF the VAS sees a significant loading reflected through the outputs and drivers.
That is my worry with your proposal. low impedance and lower current gain will stress the VAS. Simply looking at maximum current and maximum voltage specifications gets you no where near the operating conditions.
I do not mean to be doom and gloom, but you must look deeper into what you propose. Can Spice help you identify the worst combination of instantaneous current and voltage in the VAS?
MJL, you shouldn't use 2n5400, it has only 120v rating, with +/- 56v, this becomes a too close call in the vas-devices.
Use the 2n5401 instead, it has a 150v rating.
Another problem, with the increased rails, you will not be able to drive 4ohms, no matter how many output devices you are using, the VAS simply does not deliver enough current at higher frequencies.
Into 8ohms, there shouldn't be a problem.
Sadly, you can't easily increase vas-current, the compensation would need to be completely reworked and power dissipation in the to92 transistors would become unacceptable.
Some suggestions:
56ohms for your r9 is a bit large, you should use very small values here. R2/8 need to be perfectly identical, ideally even matched. 220uF for c4 is a good value. Consider 10pF for c3, this cap is critical. You might prefer a 470ohm pot for r17. r30/26 is supposed to be 47ohms. Use an extra base stopper for each output device. It's not necessary to increase q10 to bd139, it will only see ~2.4v.
An alternative to the input devices could be 2sc2240bl, having a 120v rating and hfe > 350.
Mike
Use the 2n5401 instead, it has a 150v rating.
Another problem, with the increased rails, you will not be able to drive 4ohms, no matter how many output devices you are using, the VAS simply does not deliver enough current at higher frequencies.
Into 8ohms, there shouldn't be a problem.
Sadly, you can't easily increase vas-current, the compensation would need to be completely reworked and power dissipation in the to92 transistors would become unacceptable.
Some suggestions:
56ohms for your r9 is a bit large, you should use very small values here. R2/8 need to be perfectly identical, ideally even matched. 220uF for c4 is a good value. Consider 10pF for c3, this cap is critical. You might prefer a 470ohm pot for r17. r30/26 is supposed to be 47ohms. Use an extra base stopper for each output device. It's not necessary to increase q10 to bd139, it will only see ~2.4v.
An alternative to the input devices could be 2sc2240bl, having a 120v rating and hfe > 350.
Mike
No one should confuse me with someone who knows what he's doing. I'm just tinkering here. 😉
I'm trying to get this done the poor mans way, by doing it with the parts I already have.
I have MPSA92 that I could use for Q4, Q5 and Q7. 300V but a little slow. They OK?
It really was never my intent to drive 4 ohms. I only have on hand 4 each of the output devices (MJL4281A, MJL4302A). I'll use 2 pairs instead of 3.
This is an older Toshiba part? I can't get it from Digikey (my usual source).
FWIW, I'll carry on with the 2N5550 in the input, mainly because I have these.
Any current production alternates that I could pick up for future use?
I'm trying to get this done the poor mans way, by doing it with the parts I already have.
MikeB said:
I have MPSA92 that I could use for Q4, Q5 and Q7. 300V but a little slow. They OK?
It really was never my intent to drive 4 ohms. I only have on hand 4 each of the output devices (MJL4281A, MJL4302A). I'll use 2 pairs instead of 3.
This is an older Toshiba part? I can't get it from Digikey (my usual source).
FWIW, I'll carry on with the 2N5550 in the input, mainly because I have these.
Any current production alternates that I could pick up for future use?
I am almost in the same boat as MJL21193. I have a old denon receiver with output problems, it has 400VA torroidal transformer with 52v DC rectified output. Also I have MJL4281 and MJL1302 devices. So I want to utilize these things once I receive the PCBs from Ryssen.
Also I want to reduce the overall gain of the amp to somewhere around 25, so that I can use a preamplifier. Somebody posted in thread the changes required for these output devices and reduced gain. If I implement those changes is everything will be ok? If it is not a big deal can somebody redraw schematic with correct values for onsemi devices MJL4281, MJL1302s and a gain of 25?
My second question is what should be the hfe of the input transistors 2n5551? BOM uses onsemi devices from digi-key which has minimum hfe of 80. Are they good enough? If I use these low hfe devices do I still need the emitter degeneration resistors for an amplifier of gain 25?
Otherwise mouser has fairchild devices with hfe around 250. Do we need to use them? My only concern about these fairchild devices is the middle pin is the collector so we have bent those pins to fit them in the AAK PCB.
Also I want to reduce the overall gain of the amp to somewhere around 25, so that I can use a preamplifier. Somebody posted in thread the changes required for these output devices and reduced gain. If I implement those changes is everything will be ok? If it is not a big deal can somebody redraw schematic with correct values for onsemi devices MJL4281, MJL1302s and a gain of 25?
My second question is what should be the hfe of the input transistors 2n5551? BOM uses onsemi devices from digi-key which has minimum hfe of 80. Are they good enough? If I use these low hfe devices do I still need the emitter degeneration resistors for an amplifier of gain 25?
Otherwise mouser has fairchild devices with hfe around 250. Do we need to use them? My only concern about these fairchild devices is the middle pin is the collector so we have bent those pins to fit them in the AAK PCB.
routhun said:
Naw, I'm on to better things now. Too much hassle to bend this design to my needs.
I'm playing with a "blameless" now. Sims with much less distortion.
BTW, I hope you meant MJL4302A, not MJL1302A. They don't make a complimentry pair.
Hi,
the input pair of transistors benefit from high hFE. Try for better than 400. Select by measuring Vbe when the test circuit is pushing the operating quiescent current through the collector. Try to get max difference <=1mV.
Once you have the Vbe measured then reset your test circuit to measure hFE. Match the equal Vbe transistors to better than 5%.
Now recheck the selected pair for matching Vbe before fitting to the PCB.
Take care to avoid loading the test circuit with the DMM impedance. It can screw up your readings.
the input pair of transistors benefit from high hFE. Try for better than 400. Select by measuring Vbe when the test circuit is pushing the operating quiescent current through the collector. Try to get max difference <=1mV.
Once you have the Vbe measured then reset your test circuit to measure hFE. Match the equal Vbe transistors to better than 5%.
Now recheck the selected pair for matching Vbe before fitting to the PCB.
Take care to avoid loading the test circuit with the DMM impedance. It can screw up your readings.
Zetex makes some nice TO92 transistors with high gains that make good input transistors. I recently measured some ZTX694B (npn, 120V) transistors with hFE's of 900->1000. The pnp's (ZTX795A, 140V) measured somewhat lower, but still quite high hFE (the datasheet quotes hFE's of 300-800 at 10 mA).
Hi routhun,
Hi MJL21193,
Hi Andrew,
Agreed. High gain with a linear device. Transistors that are popular for diff pair duty as used for a reason.
-Chris
Denon can make some surprisingly good amplifiers. Have you looked at the schematic of it yet? You may fine a perfectly good amp already in there.
Hi MJL21193,
This illustrates the problems of attempting to beef up a lower power design. A fresh start using some ideas from the "old" design is better than trying to pump up the older design. At least you should have learned some things of value reading these threads. The SymAsym is great in it's power range. It's better than what I expected.
Hi Andrew,
Agreed. High gain with a linear device. Transistors that are popular for diff pair duty as used for a reason.
-Chris
Hi Chris,
No I don't have the schematics of that receiver (it is Denon DRA-635r), also I am not completely tearing down the power amp section. The power amplifier section in this receiver is isolated, just I'll remove that section and replace it with SymAsym. Once I get the schematics I will fix this power amplifier also. If I like the sound I'll keep it and find another case and power supply for SymAsym.
Thanks,
routhun
No I don't have the schematics of that receiver (it is Denon DRA-635r), also I am not completely tearing down the power amp section. The power amplifier section in this receiver is isolated, just I'll remove that section and replace it with SymAsym. Once I get the schematics I will fix this power amplifier also. If I like the sound I'll keep it and find another case and power supply for SymAsym.
Thanks,
routhun
Deporting OP transistors
In the "revamp your amp" serie, I plan to give a 2nd life to a monster Hitachi HMA-8300. Those Class-E beasts designed in '77 features a hefty transformer and el-caps to match providing to pairs of supply rails (+/- 39 Vdc and +/- 95 Vdc). I will of course use the former.
I plan to fit 2 Symasym boards instead of the desperatly not repairable main board (obsolete japanese transistors) until I find a more suitable case and supply. Though the 8300 sports so nice peak meter indicators that I will perhaps change my mind and stay with it ...
To the question :
The 8300 heatsinks configuration prevents direct mounting of the output power transistors (2SA1943/2SC5200).
Is it possible to deport them with 2 pairs of 3 wires ?
Which maximum length (original 8300 OPT were 20 cm / 8" apart) ?
Which caveats (oscillation, hum, ...) ?
Which gauge (is AWG 14 a good guess) ?
Well-known designs (eg Quad 303) have OPT wired apart so I think it should not be impossible to do that ...
TIA for your advices.
Hervé
In the "revamp your amp" serie, I plan to give a 2nd life to a monster Hitachi HMA-8300. Those Class-E beasts designed in '77 features a hefty transformer and el-caps to match providing to pairs of supply rails (+/- 39 Vdc and +/- 95 Vdc). I will of course use the former.
I plan to fit 2 Symasym boards instead of the desperatly not repairable main board (obsolete japanese transistors) until I find a more suitable case and supply. Though the 8300 sports so nice peak meter indicators that I will perhaps change my mind and stay with it ...
To the question :
The 8300 heatsinks configuration prevents direct mounting of the output power transistors (2SA1943/2SC5200).
Is it possible to deport them with 2 pairs of 3 wires ?
Which maximum length (original 8300 OPT were 20 cm / 8" apart) ?
Which caveats (oscillation, hum, ...) ?
Which gauge (is AWG 14 a good guess) ?
Well-known designs (eg Quad 303) have OPT wired apart so I think it should not be impossible to do that ...
TIA for your advices.
Hervé
AAK repled:
"Hi Sheldon,
Using an external PS will work fine. Just make sure that you bypass D1 and D3, and not D2 and D4. Here's a schematic you can use for reference.
Regards,
Al"
My question again about a big central supply (well filtered) as opposed to individualdiode /caps on each board. I am starting with 4 channels and will add at least one more.
1.I see little reason to keep 2 diodes in circuit
2. there should be some decoupling capacitors although Ii think 4x4700 is huge overkill. It may be my suggested 2 x 1500ufd might not be enough.
3. 0R2 resistors as input to decoupling caps would also aid in decoupling. so I would leave these in.
Thanks,
SheldonD
.
"Hi Sheldon,
Using an external PS will work fine. Just make sure that you bypass D1 and D3, and not D2 and D4. Here's a schematic you can use for reference.
Regards,
Al"
My question again about a big central supply (well filtered) as opposed to individualdiode /caps on each board. I am starting with 4 channels and will add at least one more.
1.I see little reason to keep 2 diodes in circuit
2. there should be some decoupling capacitors although Ii think 4x4700 is huge overkill. It may be my suggested 2 x 1500ufd might not be enough.
3. 0R2 resistors as input to decoupling caps would also aid in decoupling. so I would leave these in.
Thanks,
SheldonD
.