I have started to like the 90° options…
Either mount the toroids vertically at the front, side by side if space allows, or both along the middle axis (one behind the other), place capbanks left/right accordingly, if it doesn’t fit, place them vertically? (Capbank-width < chassis-height?)
Either mount the toroids vertically at the front, side by side if space allows, or both along the middle axis (one behind the other), place capbanks left/right accordingly, if it doesn’t fit, place them vertically? (Capbank-width < chassis-height?)
For a balanced configuration, each phase sees 1/2 the speaker load and has 1/2 the voltage swing requirement. The current drive requirement remains the same. Thus in order to reduce the heatsink power dissipation you must lower the rail voltages according to your output power goals. For example, a balanced configuration capable of 25W into 8R requires a minimum of +/-10V rails and a bias current of 1.25A (or less for the Holy Grail because of the undegenerated OS).
@myleftear - Interesting... I've never considered mounting the cap banks vertically. Yes, they are < chassis height. I'll take a look and see how that might lay out. Thanks for the idea!
@lhquam - Thanks, Lynn. Totally understood. Oddly enough... I don't really have output power goals. 100W into 4R is gracious plenty. Heck, 50W is plenty. What I had before was absolutely insanely good (except in one particular use case).
The goal is to eke out what I can with what I have at the moment while still being somewhat reasonable re: configuration and not setting up your lovely amp in an absurd way that would deteriorate performance. It's counterintuitive, I know, to set up balanced / bridged when I don't need power into 8R / 16R. Parallel (with most of my speakers) is the better choice, but I'm having some fun, and I don't know if you've tested the amp in a parallel configuration.
Some back of the matchbook math leads me to think I can sneak away with 24V rails and lowering the bias a touch. I don't have any spare 12-15VAC donuts at the moment, but I may grab an extra few pair. That may be my longer-term plan.
In addition, in post #662, it's mentioned that a few minor resistor changes were needed. What were those changes, please. I've missed the details on those somewhere.
----------------------------------------------------------------------------------------------------------------------
For all -
If anyone would double check below, I'd be grateful...
I admit to not understanding fully ... but I'm pretty sure that I can't double the available current in a bridged configuration, so the standing current to drive each half of the signal would be the same... I think.
Example -
The standard is roughly 60W dissipation per board. 120W on these sinks may be pushing it.
I've run 100W per side on these sinks before with no issues with heat. So... keeping the transformers the same, and not going to any measures to lower the rails, I can adjust the bias current to control the heat.
I am clearly missing something with power calculations b/c when I reference post #679 it says...
Here are some of the parameters:
What I wind up with using 24V rails and an Iq for each board of 1A04 for each board is below. Help with my wee brain's misunderstanding is greatly appreciated.
As an example... when I plug Lynn's numbers into my spreadsheet, I get below. I am clearly messing something up (per usual) with the current delivery.
I have zero, zip, nada concerns with the potential power delivery. This is an exercise for me to learn. I just can't wrap my brain around the current delivery in balanced into the load.
My gratitude in advance for anyone that can dumb it down for me. I promise I've tried... and read... then read more.
@lhquam - Thanks, Lynn. Totally understood. Oddly enough... I don't really have output power goals. 100W into 4R is gracious plenty. Heck, 50W is plenty. What I had before was absolutely insanely good (except in one particular use case).
The goal is to eke out what I can with what I have at the moment while still being somewhat reasonable re: configuration and not setting up your lovely amp in an absurd way that would deteriorate performance. It's counterintuitive, I know, to set up balanced / bridged when I don't need power into 8R / 16R. Parallel (with most of my speakers) is the better choice, but I'm having some fun, and I don't know if you've tested the amp in a parallel configuration.
Some back of the matchbook math leads me to think I can sneak away with 24V rails and lowering the bias a touch. I don't have any spare 12-15VAC donuts at the moment, but I may grab an extra few pair. That may be my longer-term plan.
In addition, in post #662, it's mentioned that a few minor resistor changes were needed. What were those changes, please. I've missed the details on those somewhere.
----------------------------------------------------------------------------------------------------------------------
For all -
If anyone would double check below, I'd be grateful...
I admit to not understanding fully ... but I'm pretty sure that I can't double the available current in a bridged configuration, so the standing current to drive each half of the signal would be the same... I think.
Example -
The standard is roughly 60W dissipation per board. 120W on these sinks may be pushing it.
| Target Dissipation Per Board (W) | Rails (+ / -) | Target Current | Voltage Across Sense Resistors (0R11) |
60 | 24 | 1.25 | 0.138 |
I've run 100W per side on these sinks before with no issues with heat. So... keeping the transformers the same, and not going to any measures to lower the rails, I can adjust the bias current to control the heat.
| Target Dissipation Per Board (W) | Rails (+ / -) | Target Current | Voltage Across Sense Resistors (0R11) |
50 | 24 | 1.04 | 0.115 |
I am clearly missing something with power calculations b/c when I reference post #679 it says...
Here are some of the parameters:
- Rail Voltages: +/-29V. Max voltage for the MeanWell LRS-200N2-24 SMPS.
- Bias current: 1.18A
- Heatsink Temnperature: Stable at 27C above ambient.
- FETs: FQA28N15 / FQA36P15
- 8 Ohms Load: output is rail voltage limited to 170 Watts (+/-52V peak) with a THD of 0.01%.
- 4 Ohms load: harmonic rolloff clean until around 240 Watts (+/-43.8V peak) with THD of 0.03%.
What I wind up with using 24V rails and an Iq for each board of 1A04 for each board is below. Help with my wee brain's misunderstanding is greatly appreciated.
| Bias Current Per Board | Max Current "Swing" In Balanced (A) | Max Voltage Swing (Vpp) In Balanced | Vrms | Load | Max Power Output - Voltage Limit (V^2/R) | Max Output - Current Limit (R*I^2) |
1.04 | 4.17 | 96 | 33.9411255 | 16 | 72 | 278 |
1.04 | 4.17 | 96 | 33.9411255 | 8 | 144 | 139 |
1.04 | 4.17 | 96 | 33.9411255 | 6 | 192 | 104 |
1.04 | 4.17 | 96 | 33.9411255 | 4 | 288 | 69 |
As an example... when I plug Lynn's numbers into my spreadsheet, I get below. I am clearly messing something up (per usual) with the current delivery.
| Target Dissipation Per Board (W) | Rails (+ / -) | Target Current | Voltage Across Sense Resistors (0R11) | |||
60 | 26 | 1.18 | 0.130 | |||
| Bias Current Per Board | Max Current "Swing" In Balanced (A) | Max Voltage Swing In Balanced | Vrms | Load | Max Power Output - Voltage Limit (V^2/R) | Max Output - Current Limit (R*I^2) |
1.18 | 4.72 | 104 | 36.76955262 | 16 | 85 | 356 |
1.18 | 4.72 | 104 | 36.76955262 | 8 | 169 | 178 |
1.18 | 4.72 | 104 | 36.76955262 | 6 | 225 | 134 |
1.18 | 4.72 | 104 | 36.76955262 | 4 | 338 | 89 |
I have zero, zip, nada concerns with the potential power delivery. This is an exercise for me to learn. I just can't wrap my brain around the current delivery in balanced into the load.
My gratitude in advance for anyone that can dumb it down for me. I promise I've tried... and read... then read more.
^ Replying to myself... and hoping that I may have just realized my error after posting... per normal. As always, just when I'm at the end of my rope and reach out for help... I somehow get another idea. 
Above would be power delivery in CLASS A, correct?!
The amp can source more current up to ?? in Class A/B, correct?!
So, how do we know the current limit in A/B?
Edited to add - I'm also considering putting in four LRS-200N2-24 SMPS that Lynn recommended. That would solve my weight / space issues pretty much instantly. 🙂
Above would be power delivery in CLASS A, correct?!
The amp can source more current up to ?? in Class A/B, correct?!
So, how do we know the current limit in A/B?
Edited to add - I'm also considering putting in four LRS-200N2-24 SMPS that Lynn recommended. That would solve my weight / space issues pretty much instantly. 🙂
Last edited:
You divide the rail voltage - mosfet gate voltage to half of the load impedance.
For 20v rails and 4r load you will have ~ (20-5)/2 which is 7.5a
Big boys trigger the ops protection at 10a on the pucks.
Is there any impact on distortion by lowering the rails to say +/-12V for those Fairchild MOSFETs ?
Patrick
I'd have never figured that out in a million years... much appreciated!
Copied into my learning notes to read a bit more and determine how this works vs. just knowing the answer.
For now... carrying on.
Both 'new' boards have been tested on the bench. They bias up properly, and offset is nulled. Popping them into the chassis in place of the previous boards for a full check.
Then, PSU decisions...
Much appreciated to you (and everyone) for the assistance and pointing toward more things to learn.
Here’s how I understand it. If you have 20V rails, then max swing is 20 - Vgs, about 16V, and 16V peak into 4R is 64W peak (P=V^2 / R). You’d clip the rails beyond that. Is that right?
If you do it right, Vgs can swing beyond the rails.
Buy there is always need for some Vds.
It varies between devices.
Patrick
Buy there is always need for some Vds.
It varies between devices.
Patrick
So... I know this thread isn't 'Teach Patrick amplifier basics', so I'll keep it short(er) in the hopes that it may help others.
If I have it correct... then below would be roughly the theoretical Class A power output in Green with 24V rails in a balanced configuration.
And below would be the theoretical maximum power output based on the additional current ... from (at this point it may as well be magic to me)...
So for the current delivery... even though it doesn't really matter in this particular situation, is it the Class A current OR the "other calculation" whichever is greatest?
Example - in this case, the rails limit the output into 16R anyway, but the current from the calculation is < the Iq.
Clearly there are some other factors at play that I don't need to worry about at the moment, because the theoretical power output doesn't quite match Lynn's. However, it seems I'm getting warmer. Heck, even the Class A power delivery should be more than sufficient. Knowing a bit of the background re: why I get the additional headroom into lower impedance loads is fascinating. More to learn.
Again... truly appreciated re: the patience to teach.
Edited to add - I just noticed that the 'Real' Patrick posted just prior. Clearly they don't need to learn amplifier basics. I do.
Cheers,
(Another) Patrick
If I have it correct... then below would be roughly the theoretical Class A power output in Green with 24V rails in a balanced configuration.
| Bias Current Per Board | Max Current "Swing" In Balanced (Class A) | Max Voltage Swing In Balanced | Vrms | Load | Max Class A Power Output - Voltage Limit (V^2/R) | Max Class A Output - Current Limit (R*I^2) |
1.04 | 4.17 | 96 | 34 | 16 | 72 | 278 |
1.04 | 4.17 | 96 | 34 | 8 | 144 | 139 |
1.04 | 4.17 | 96 | 34 | 6 | 192 | 104 |
1.04 | 4.17 | 96 | 34 | 4 | 288 | 69 |
And below would be the theoretical maximum power output based on the additional current ... from (at this point it may as well be magic to me)...
| Bias Current Per Board | Max Current "Swing" In Balanced (A/B) | Max Voltage Swing In Balanced | Vrms | Load | MOSFET Gate Voltage | Max Power Output - Voltage Limit (V^2/R) | Max Output - Current Limit (R*I^2) |
1.04 | 2.38 | 96 | 33.9411255 | 16 | 5 | 72 | 90 |
1.04 | 4.75 | 96 | 33.9411255 | 8 | 5 | 144 | 181 |
1.04 | 6.33 | 96 | 33.9411255 | 6 | 5 | 192 | 241 |
1.04 | 9.50 | 96 | 33.9411255 | 4 | 5 | 288 | 361 |
So for the current delivery... even though it doesn't really matter in this particular situation, is it the Class A current OR the "other calculation" whichever is greatest?
Example - in this case, the rails limit the output into 16R anyway, but the current from the calculation is < the Iq.
Clearly there are some other factors at play that I don't need to worry about at the moment, because the theoretical power output doesn't quite match Lynn's. However, it seems I'm getting warmer. Heck, even the Class A power delivery should be more than sufficient. Knowing a bit of the background re: why I get the additional headroom into lower impedance loads is fascinating. More to learn.
Again... truly appreciated re: the patience to teach.
Edited to add -
I just noticed that the 'Real' Patrick posted just prior. Clearly they don't need to learn amplifier basics. I do.Cheers,
(Another) Patrick
Here‘s the first time I didit: https://www.diyaudio.com/community/attachments/topview-jpg.1011710/
And here the secont time, this the HGF (you may remember it): https://www.diyaudio.com/community/attachments/img_6313-jpeg.1310440/
And here the secont time, this the HGF (you may remember it): https://www.diyaudio.com/community/attachments/img_6313-jpeg.1310440/
The Ciss, Crss, Coss capacitances will be higher as Vds is lowered. Some even harmonics due to that increase will be canceled in the balanced amplifier.
@lhquam - Hi Lynn - I hate to bug you with trivial questions, but my earlier question may have gotten lost in the wall of text. What resistor changes are required? I'm referencing below from post #662 (emphasis added is mine).
Is this proper for wiring? Same for both left and right channels with regard to the ground break?
Once again, I appreciate all the help. I should probably just wire it up and learn, but the entire concept of the ground break on one board has me admittedly baffled, and the chassis portion of this one... has me wanting desperately to get it done properly the first time and leave it alone.
With my thanks!
Edited to remove the photo with the incorrect RCA wiring. I don't intend to use RCA. Grabbed the wrong file, and the wiring was wrong.
The wiring seems intuitive and straight forward. Where I once again need some assistance is with the ground loop breaking amplifier channel(s).
Is this proper for wiring? Same for both left and right channels with regard to the ground break?
Once again, I appreciate all the help. I should probably just wire it up and learn, but the entire concept of the ground break on one board has me admittedly baffled, and the chassis portion of this one... has me wanting desperately to get it done properly the first time and leave it alone.
With my thanks!
Edited to remove the photo with the incorrect RCA wiring. I don't intend to use RCA. Grabbed the wrong file, and the wiring was wrong.
I have operated the ZD25-Mini with rail voltages between 22V and 29V without resistor changes, other than ones outlined in the Build Guide that "tune" the bias circuit.
^ Wonderful!!!! Thank you. I am almost ready to fire up the additional channels for music testing. Then I can wire them in the balanced configuration.
With no comment or
for my crude wiring diagram, I'll assume that won't blow anything up.
With no comment or
for my crude wiring diagram, I'll assume that won't blow anything up.
Early success!
PSUs came yesterday. I selectively omitted from my brain that they only count 'business days' for delivery. So, I didn't get a chance to work on it over the weekend.
Initial / crude settings:
~20V8 rails (as low as the PSUs would go)
0V130 across sense resistors => 1A2 Iq
Max offset 'within' board - 1.2mV
Max 'overall offset' at speaker terminals between boards - 2.8mV
Dissipation per side ~ 100W (50 per sink). Based on temps, I'll see where I want to move voltage and/or bias current.
Ignore the spectacular display of wire management... Oddly enough... it's dead quiet though.
I'll get the final settings tweaked and tidy up wiring. Then... it should be in the system for a long time.
I'll attempt some measurements toward the weekend maybe.
@lhquam - THANK YOU!
Edited to add - This thing is MUCH easier on my back... 🙂
Edited again to correct dissipation - I was doing it per sink / per side.
PSUs came yesterday. I selectively omitted from my brain that they only count 'business days' for delivery. So, I didn't get a chance to work on it over the weekend.
Initial / crude settings:
~20V8 rails (as low as the PSUs would go)
0V130 across sense resistors => 1A2 Iq
Max offset 'within' board - 1.2mV
Max 'overall offset' at speaker terminals between boards - 2.8mV
Dissipation per side ~ 100W (50 per sink). Based on temps, I'll see where I want to move voltage and/or bias current.
Ignore the spectacular display of wire management... Oddly enough... it's dead quiet though.
I'll get the final settings tweaked and tidy up wiring. Then... it should be in the system for a long time.
I'll attempt some measurements toward the weekend maybe.
@lhquam - THANK YOU!
Edited to add - This thing is MUCH easier on my back... 🙂
Edited again to correct dissipation - I was doing it per sink / per side.
^ Perfectly acceptable.
Max ~50C on the hottest place on the sinks between the fins and ~40C on the tips of the sinks. Measured with the el-cheapo K-type probes that come with just about every DMM. Also agrees with the fact that they are quite warm, but not "HOT". I could hold onto them all day.
22-23C ambient => 27-28C rise over ambient.
I estimate 0.30C per Watt for my Modushop 5U/400 as a conservative starting point for the hottest point and about 0.20C / W for the fins. I'm usually a tad under that, but I like to err on the side of caution.
I base most of my estimates on a really rough regression I did based on heatsink dimensions and the study that Modushop published on a few of their chassis. I've got a spreadsheet, and I just input the different dimensions ... and it gets me in the right neighborhood. A more simple way (that also works well) is to just know that it will fall somewhere between the published information from the 4U/400 and 5U/500. My chart is below. I pulled the data visually from the study published by Modushop. Their ambient was a touch higher than my typical if I recall... so... YMMV. I'm typically off by a few degrees.
Max ~50C on the hottest place on the sinks between the fins and ~40C on the tips of the sinks. Measured with the el-cheapo K-type probes that come with just about every DMM. Also agrees with the fact that they are quite warm, but not "HOT". I could hold onto them all day.
22-23C ambient => 27-28C rise over ambient.
I estimate 0.30C per Watt for my Modushop 5U/400 as a conservative starting point for the hottest point and about 0.20C / W for the fins. I'm usually a tad under that, but I like to err on the side of caution.
I base most of my estimates on a really rough regression I did based on heatsink dimensions and the study that Modushop published on a few of their chassis. I've got a spreadsheet, and I just input the different dimensions ... and it gets me in the right neighborhood. A more simple way (that also works well) is to just know that it will fall somewhere between the published information from the 4U/400 and 5U/500. My chart is below. I pulled the data visually from the study published by Modushop. Their ambient was a touch higher than my typical if I recall... so... YMMV. I'm typically off by a few degrees.