[f] AMB on Facebook

THE BALANCED BETA 22 BUILD LOG

beta22

Re: THE BALANCED BETA 22 BUILD LOG

Postby amb » September 24th, 2019, 4:19 pm

Switching around the power Supplies is a good first step to isolating the problem. If that doesn’t move the noise to the other channel, you can take the process one step further by physically swapping the amp boards from side to side. If the problem moves to the other side, then you’ve identified a problem with at least one of the boards. If the noise stays in the same channel, then the problem is external to the boards you moved.

Energy likes this post.
User avatar
amb
Site Admin
 
Posts: 10382
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby Energy » September 25th, 2019, 1:28 am

I wasn't aware that you merged my two threads together so in the last post that I worked so hard to create I edited it off and saved without knowing. Is there any way that I can restore it?
Computer (Ethernet) Intel NUC7i7DNHE (USB) SOtM tX-USBultra (USB) Denafrips Terminator (XLR) AMB Laboratories β22 (Balanced) Focal Utopia

= ︿ =
User avatar
Energy
 
Posts: 86
Likes: 3 posts
Liked in: 4 posts
Joined: September 30th, 2012, 4:07 pm
Location: San Francisco, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby amb » September 25th, 2019, 6:34 am

Energy wrote:I wasn't aware that you merged my two threads together so in the last post that I worked so hard to create I edited it off and saved without knowing. Is there any way that I can restore it?

I merged your two topics because they had the exact same titles, but I don't know what happened that would cause a post to be lost. What you see is what there is, if something disappeared, it's gone. You'll have to re-create it. Sorry.

Energy likes this post.
User avatar
amb
Site Admin
 
Posts: 10382
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby Energy » September 26th, 2019, 12:39 am

all good, thanks for trying. Is it cool if i asked a couple of questions on the B22?

1. Is the 2W 5% metal-oxide resistor 0.47Ω a required part of the output? In other words, is the resistance important?
If I were to substitute it for a 2W 5% metal film would the small inductance from metal film have a significant degradation to the sound quality?
I know It's used to heat up in order to protect the MOSFETS from shorting, but if one uses a balanced configuration which doesn't have the temporary short circuits that comes from plugging TRS plugs and wanted to bypass it, would that be possible or would it still require the 0.47Ω impedance? Under normal operations it should be good, and I am fine with potentially damaging a MOSFET (it can be fixed) if a resistor can be removed. Furthermore, is the reason why you don't suggest going above 2W due to wanting the Oxide heat up and burn first before the components do? I'm guessing 3W and above will damage the nearby components instead of the standalone resistor.

2. A similar question for the 2W 5% metal-oxide resistor 22Ω in the Zobel network as well. Would metal film, given the lower noise still not be as good given the inductance? Or does the Metal Oxide play better in the ultrasonic frequencies?

3. For the zobel wouldn't a 0.047µF C0G/NPO perform better than the specified metallized polyester or X7R that was given on the parts sheet? (lower ESR, better ultrasonic's, more stable)

4. Would a metal foil trimpot have lower noise figures than a cermet trimpot due to better resistive element?

4. The LSK170B is slightly above the 2SK370BL in terms of noise right?

5. Would there be any theoritical benefits to matching the zener and signal diodes to have the same forwarding current?

6. Is it possible to use slightly higher capacitance capacitors (270uF instead of 100uF, 820uF instead of 470uF) for the sake of lowering ESR?

7. Is it most optimal if the Q1-Q4 for all four Beta 22 boards had the same Idss to one another? Or is after the adjustments on trimpots does it not really matter as much?

8. If I used much larger heatsinks, is it possible to set the 120mA-160mA quiescent current on VRM2 Trimpot to say 180mA or 200mA for more output swing?
Would there be any other areas to increase aside from this if it were possible (eg. VRM 1's 4.5V).

9. What is the recommended gain for a balanced beta 22 or the one suggested for the best sound quality?

10. Is the ε24 power switch driver circuit and σ24 power transformer board the only other two things required to use a Bulgin MPI002?

11. Are the amplifier boards that sensitive to one another? The circuits are facing one another while the heatsinks are in opposite direction. Would there be a SNR or sonic benefit to space them out further? I had my last one slightly stacked on the other and wonder if extra distance would help.

12. For the Sigma 22's 3900pF, can a 3900pF C0G/NP0 be used?

13. Is the EMi/RFI receptacle really that needed? Is the sigma's PSU sensitive to that or is there an actual gain with using one? I was planning on using a direct IEC power entry receptacle but wondered if the previous was still worth keeping.

14. Outside of requiring four Sigma 22's to drive speakers, is there any benefit to using four separate transformers/Sigma 22's for headphone amplification? Does having individual transformers and Sigma 22's for each of the four channels help with noise or adds additional filtering and isolation? Is the recommended VA for speaker amplification 100VA? I will be beefing up the heatsink, like question #8, is there a way to increase the 50W balanced output rating if I raised the trimpots? Would increasing the 100VA to say 120VA give more headroom or would it start to become restricted by the Sigma 22's circuitry.

15. For headphone amplification alone, balanced configuration, would two smaller 60VA's per Sigma 22 be better for noise? In terms of not leaking magnetic waves as much if the unit were made to be stacked.
Computer (Ethernet) Intel NUC7i7DNHE (USB) SOtM tX-USBultra (USB) Denafrips Terminator (XLR) AMB Laboratories β22 (Balanced) Focal Utopia

= ︿ =
User avatar
Energy
 
Posts: 86
Likes: 3 posts
Liked in: 4 posts
Joined: September 30th, 2012, 4:07 pm
Location: San Francisco, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby amb » September 26th, 2019, 11:56 am

Energy wrote:1. Is the 2W 5% metal-oxide resistor 0.47Ω a required part of the output? In other words, is the resistance important?

The output resistors serve at least two important functions:
- Output MOSFET protection
- Enhance thermal stability for the MOSFETs

If I were to substitute it for a 2W 5% metal film would the small inductance from metal film have a significant degradation to the sound quality?

No, but I specify metal oxide type over others for its unique characteristic that when overloaded, the resistance increases (and eventually goes open circuit like a fuse). This helps with the protection aspect. Other resistor types tend to short circuit as they "blow", making matters worse.

I know It's used to heat up in order to protect the MOSFETS from shorting, but if one uses a balanced configuration which doesn't have the temporary short circuits that comes from plugging TRS plugs and wanted to bypass it, would that be possible or would it still require the 0.47Ω impedance?

You should never install a short circuit or otherwise bypass these resistors. That would not provide any benefit anyway. 0.47Ω is specified as the minimum.

Under normal operations it should be good, and I am fine with potentially damaging a MOSFET (it can be fixed) if a resistor can be removed.

Why? Do you somehow believe that you would gain something by doing that?

Furthermore, is the reason why you don't suggest going above 2W due to wanting the Oxide heat up and burn first before the components do? I'm guessing 3W and above will damage the nearby components instead of the standalone resistor.

Correct. Even 2W is very generous for headphones.

2. A similar question for the 2W 5% metal-oxide resistor 22Ω in the Zobel network as well. Would metal film, given the lower noise still not be as good given the inductance? Or does the Metal Oxide play better in the ultrasonic frequencies?

I specify metal oxide here for the same reason, because metal oxide does not short circuit when it burns up. We don't want a shorted resistor at the output of the amp, even if it's in series with a capacitor.

3. For the zobel wouldn't a 0.047µF C0G/NPO perform better than the specified metallized polyester or X7R that was given on the parts sheet? (lower ESR, better ultrasonic's, more stable)

You can use a multilayer ceramic capacitor if you want, but typically they are very small and do not lend themselves to being "free air" mounted as well as a larger film capacitor. For the purpose of this capacitor, metallized polyester is just fine.

4. Would a metal foil trimpot have lower noise figures than a cermet trimpot due to better resistive element?

Noise is not really an issue here. I don't know which metal foil trimpot you have in mind, but the cermet type specified works just fine...

4. The LSK170B is slightly above the 2SK370BL in terms of noise right?

No. They're the same.

5. Would there be any theoritical benefits to matching the zener and signal diodes to have the same forwarding current?

No. And I think you're referring to the zener voltage, not current.

6. Is it possible to use slightly higher capacitance capacitors (270uF instead of 100uF, 820uF instead of 470uF) for the sake of lowering ESR?

Be careful, too much capacitance can be counter-productive. The startup charge current will be increased, stressing other parts.

7. Is it most optimal if the Q1-Q4 for all four Beta 22 boards had the same Idss to one another? Or is after the adjustments on trimpots does it not really matter as much?

No, each board is its own entity. You do not need to match Idss across boards. Althogh typically all JFET quads I ship from AMB audio shop (within each order) are matched, but it's not a guarantee.

8. If I used much larger heatsinks, is it possible to set the 120mA-160mA quiescent current on VRM2 Trimpot to say 180mA or 200mA for more output swing?
Would there be any other areas to increase aside from this if it were possible (eg. VRM 1's 4.5V).

Increasing the quiescent does not increase output voltage swing. The swing is limited by the supply voltage.

9. What is the recommended gain for a balanced beta 22 or the one suggested for the best sound quality?

That's the wrong question. The correct one is to ask what is the most appropriate gain for your source/headphone combination so you don't hear the noise floor and have a good range of volume control.

10. Is the ε24 power switch driver circuit and σ24 power transformer board the only other two things required to use a Bulgin MPI002?

If you have the momentary pushbutton version of the Bulgin, yes. If you can't find an appropriate EI-30 transformer (they are easy to find in Europe, but less so in the US), then skip the σ24 and use a small chassis-mount toroid instead.

11. Are the amplifier boards that sensitive to one another? The circuits are facing one another while the heatsinks are in opposite direction. Would there be a SNR or sonic benefit to space them out further? I had my last one slightly stacked on the other and wonder if extra distance would help.

Board spacing is not that important as long as you have short wire runs, and keep them away from the power transformers.

12. For the Sigma 22's 3900pF, can a 3900pF C0G/NP0 be used?

Yes, but there is nothing wrong with the film cap specified.

13. Is the EMi/RFI receptacle really that needed? Is the sigma's PSU sensitive to that or is there an actual gain with using one? I was planning on using a direct IEC power entry receptacle but wondered if the previous was still worth keeping.

No, it's optional. Unless your AC supply is really bad in quality...

14. Outside of requiring four Sigma 22's to drive speakers, is there any benefit to using four separate transformers/Sigma 22's for headphone amplification? Does having individual transformers and Sigma 22's for each of the four channels help with noise or adds additional filtering and isolation? Is the recommended VA for speaker amplification 100VA?

For headphones only you don't need four transformers. Two is enough (for two σ22s), and 60-80VA is plenty.

I will be beefing up the heatsink, like question #8, is there a way to increase the 50W balanced output rating if I raised the trimpots? Would increasing the 100VA to say 120VA give more headroom or would it start to become restricted by the Sigma 22's circuitry.

No, as I said, the output voltage swing (and there fore the power output given a headphone load) is completely determined by the supply voltage. No amount of trimpot tweaking will increase that. That said, a balanced β22 is capable of swinging a lot of voltage, much more than any dynamic headphones will ever need.

15. For headphone amplification alone, balanced configuration, would two smaller 60VA's per Sigma 22 be better for noise? In terms of not leaking magnetic waves as much if the unit were made to be stacked.

Yes, smaller transformer = less magnetic field leakage, all else being equal. You can specify shielded transformers for even lower magnetic fields.
User avatar
amb
Site Admin
 
Posts: 10382
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby Energy » September 27th, 2019, 2:07 pm

Thanks for being so patient with the questions.

You say the 2W is generous for headphones. Would it still be set to 2W even if using the amplifier as a speaker amp? I plan on having (4) sigma 22 and (4) beta 22's, each sigma having it's own toroidal transformer. If I used 60VA x 4, would that be enough for the 50W into 8Ω balanced specification or would I require 80VA or 100VA?

Furthermore, with each of the four Sigma 22 boards having it's own transformer, each beta 22 board can now have it's own sigma 22/transformer. Does this offer more isolation between the channels or does it not matter? (since two beta 22 boards are no longer sharing the same output of one sigma 22).

So even though it's balanced, with the differences in Idss, resistor impedance, etc it still won't affect channel matching?

Could you explain the MOSFET quiescent current and why increasing it increases the MOSFET's heat? If there is adequate cooling, is there nothing to gain from increasing beyond 160mA even if driving speakers?
Computer (Ethernet) Intel NUC7i7DNHE (USB) SOtM tX-USBultra (USB) Denafrips Terminator (XLR) AMB Laboratories β22 (Balanced) Focal Utopia

= ︿ =
User avatar
Energy
 
Posts: 86
Likes: 3 posts
Liked in: 4 posts
Joined: September 30th, 2012, 4:07 pm
Location: San Francisco, CA
Country: United States (us)

THE BALANCED BETA 22 BUILD LOG

Postby amb » September 28th, 2019, 11:07 am

Energy wrote:You say the 2W is generous for headphones. Would it still be set to 2W even if using the amplifier as a speaker amp?

Yes.

I plan on having (4) sigma 22 and (4) beta 22's, each sigma having it's own toroidal transformer. If I used 60VA x 4, would that be enough for the 50W into 8Ω balanced specification or would I require 80VA or 100VA?

I have mentioned this elsewhere before, the β22 is primarily designed to be a headphone amplifier. It happens to be stout enough to drive speakers, but the output MOSFETs are not big enough (nor are the default heatsinks big enough) to make it a serious speaker amplifier. For speaker use there is the β24. The β22 should only be considered a "light duty" speaker amp, despite the possibility of making it output 50W into 8Ω, just not on a sustained basis or you may overheat and damage it. With that in mind, you need not go overboard with the power supply. 60-80VA is enough for each of the transformers.

Furthermore, with each of the four Sigma 22 boards having it's own transformer, each beta 22 board can now have it's own sigma 22/transformer. Does this offer more isolation between the channels or does it not matter? (since two beta 22 boards are no longer sharing the same output of one sigma 22).

Having four σ22s does not "isolate the stereo channels" any better than just two σ22s, whether you use 2 or 4 trandformers. The use of 4 transformers and 4 σ22s is to reduce the current load on each one when the amp is driving speakers, so that we don't overheat and damage the MOSFETs on the σ22s too.

So even though it's balanced, with the differences in Idss, resistor impedance, etc it still won't affect channel matching?

For Idss, once matched within one board, the board will function optimally. The other boards don't have to be exactly the same, because this does not affect gain, distortion, or other parameters that may make one board perform differently than another. The same cannot be said for resistors! Resistor values should match between boards, especially R3 and R4 because these affect the voltage gain. But the use of 1% resistors is good enough.

Could you explain the MOSFET quiescent current and why increasing it increases the MOSFET's heat? If there is adequate cooling, is there nothing to gain from increasing beyond 160mA even if driving speakers?

The heat dissipation of a device (be it MOSFET, BJT, diode or resistor) in Watts is the current through the device multiplied by the voltage across the device. P = I * V. If you increase I, P will also increase.

At 160mA quiescent, it is deep class A biasing for just about any headphone. But for speakers, it will fall out of class A and become class AB when the peak output current exceeds twice the amp's quiescent current. For the β22, there is no way to bias the amp high enough to class A at full speaker output power. Too much current and heat.
User avatar
amb
Site Admin
 
Posts: 10382
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby Energy » October 3rd, 2019, 1:52 pm

Thanks Ti. I realized building a α10 Balanced Pre-Amplifier with β24 in pure Class-A would be more ideal. I didn't realize that the 50W into 8Ω takes it away from Class-A and puts it into Class-AB operation.

I read somewhere on the forum that you said these MOSFETs can sound better with higher quiescent currents. I know 120-160mA should be enough to stay in Class-A for most headphones, even the inefficient ones, but would it help with the sound if I increased it to.. say 200mA given that heat wasn't a problem? Would that be the only thing I increase, nothing else?

On a side note, I noticed C9 and C10 are not shunted with a 1uF Film for lower ESR. Is there a reason why?
Computer (Ethernet) Intel NUC7i7DNHE (USB) SOtM tX-USBultra (USB) Denafrips Terminator (XLR) AMB Laboratories β22 (Balanced) Focal Utopia

= ︿ =
User avatar
Energy
 
Posts: 86
Likes: 3 posts
Liked in: 4 posts
Joined: September 30th, 2012, 4:07 pm
Location: San Francisco, CA
Country: United States (us)

Re: THE BALANCED BETA 22 BUILD LOG

Postby amb » October 3rd, 2019, 2:32 pm

Raising the quiescent current changes the point where the amp would transition from class A operation to class AB. You should not do this too aggressively to keep the MOSFETs from overheating.

C9/C10 are not bypassed with film caps because it’s not necessary. That portion of the circuit is a capacitance multiplier. The bypassing is done with C13/C14.
User avatar
amb
Site Admin
 
Posts: 10382
Likes: 175 posts
Liked in: 482 posts
Joined: March 29th, 2010, 9:49 pm
Location: Sunnyvale, CA
Country: United States (us)

Previous

Return to β22 amplifier

Who is online

Users browsing this forum: No registered users and 0 guests

Quick Links

AMB audio DIY resources
AMB audio shop

cron