||MYWAY|| 50W Class-A with Direct heatsink mounted devices

Where we discuss new analog design ideas for Pro Audio and modern spins on vintage ones.
Kanwar
Posts: 8
Joined: Fri Nov 29, 2013 8:15 am

||MYWAY|| 50W Class-A with Direct heatsink mounted devices

Post by Kanwar »

Hello everyone,

I have conceptualized and designed a CLASS-A AMPLIFIER 50W with following features.

:ugeek:

1. The mosfet devices are directly mounted on the heatsink without any insulator under it making the thermal performance simply top-notch. The output topology is Grounded Drain. Therefore the heatsink is not at any potential but Grounded.

2. The thermal Bias and temperature tracking are also mounted directly on heatsink without any insulators thereby enabling fast thermal feedback and better bias control.

3. The amplifier has a minimalist design concept. The input features pure complementary class-A buffer which accepts the input signal as well as acts as a summing node for the feedback signal. The grounded drain output topology is configured as compound feedback pair with bipolar level shifters + high speed buffers + mosfets, This type of arrangement results in linear output stage because of local feedback and better bias stability.

4. The PSRR which is important aspect in determining the ability of the amplifier in rejecting the noise from power supply is modest at 60dB which provides a typical HUM free operation with the sensitive speakers. The power supply features CRC filter arrangement having 8 X 15000MFD/65V custom made capacitors.

5. Protections included DC offset, Soft-start, Delayed Turn-on, Loss of AC detection, Over-temperature and Over Current.

The only drawback is the amount of heat it produces while giving out wattage 50W into 8 ohm load, the heat dissipation is around 200W continuous.

Conceptual schematic
Image

The initial prototype
Image

10khz Sinewave response
Image

Soft clipping
Image

10Khz Squarewave response
Image

10Khz Triangle wave
Image

100khz Square wave
Image

These above tests were conducted at rated load i.e. 8ohms

The heatsinks
Image

Image

The PCB
Image

The Final prototype
Image

Image



Cheers,
Kanwar
Last edited by Kanwar on Wed Dec 17, 2014 6:33 am, edited 1 time in total.
User avatar
mediatechnology
Posts: 5444
Joined: Sat Aug 11, 2007 2:34 pm
Location: Oak Cliff, Texas
Contact:

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by mediatechnology »

Very, very nice! Thanks for contributing. You've spent a lot of time on this.

The topology reminds me of Cherry's figure 1d:

Image

Are you going to offer kits?
Kanwar
Posts: 8
Joined: Fri Nov 29, 2013 8:15 am

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by Kanwar »

mediatechnology wrote:Very, very nice! Thanks for contributing. You've spent a lot of time on this.

The topology reminds me of Cherry's figure 1d:

Are you going to offer kits?
Thank you for the compliments, Wayne.

Yes the figure 1d: is grounded emitter topology and mine is similar in a way which is grounded drain which keeps the metal tabs[Drain/Collector] of mosfets directly mated to heatsink, i.e. thermally and electrically and the power supply rails are floating.

Surely, I can provide the kits or Pcbs. Need to think on how to do that, if anyone is interested.
emrr
Posts: 570
Joined: Sun Mar 02, 2008 10:27 pm
Location: NC, USA
Contact:

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by emrr »

Agreed, nice work here. Is there any kit like this currently? I seldom look at power amp DIY to know one way or the other. Probably worth exploring if it's not represented in the market.
Best,

Doug Williams
Electromagnetic Radiation Recorders
Kanwar
Posts: 8
Joined: Fri Nov 29, 2013 8:15 am

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by Kanwar »

emrr wrote:Agreed, nice work here. Is there any kit like this currently? I seldom look at power amp DIY to know one way or the other. Probably worth exploring if it's not represented in the market.
Thanks emrr,

Except this, there is no other class-A design/kit like this anywhere in the whole world which is having all the devices[Power transistors, Bias transistors, Thermal switches] which are meant to be mounted on heatsink are mounted without an insulator. The devices are directly mounted on heatsink to ensure ultimate heat transfer. That is the uniqueness of this design along with a very simple complementary frontend. This design is flexible in providing class-A power from 25W To 50W and in Class-AB it can do 400W @8ohms. I did it out of fun suddenly and result was good.... 8-)
Last edited by Kanwar on Wed Dec 17, 2014 10:31 pm, edited 1 time in total.
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by JR. »

The "driven rail" topology is widely done in low cost designs because the front end can be driven from low voltage rails. There can be issues trying to make a two channel amp since they need separate secondary windings. While I never did one myself I vaguely recall some speed issues associated with swinging the transformer winding with the full audio signal, bot most of these I saw were medium to lo-fi. I can't count how many millions of driven rail amp channels were sold by Peavey mostly to top boxes or PA heads. .

IIRC QSC did a version with the collectors of a bipolar device amp connected to a grounded heat sink. As I recall there may have been some issues with sensing device current for degeneration or current limiting. QSC had a patent for their current limiting approach that apparently worked, but I never felt it was worth copying.

Looks like a nice "winter" amp... I momentarily though about doing a class A variant amp, the last time I came up with a different way but nah.. too busy messing with thermostats and water filters.. and I rarely even listen to music (I got spoiled by having live bands play in my living room, the hifi never sounded the same after that).

JR
Cancel the "cancel culture", do not support mob hatred.
Kanwar
Posts: 8
Joined: Fri Nov 29, 2013 8:15 am

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by Kanwar »

JR. wrote:The "driven rail" topology is widely done in low cost designs because the front end can be driven from low voltage rails. There can be issues trying to make a two channel amp since they need separate secondary windings.
Yes, it is low cost because the frontend is easy to construct and low cost doesn't always means a cheap design. Having a floating rails design it requires a separate power supply for each channel, I don't see any problem in that. By the way dual mono amplifiers are always preferred instead of stereo amp sharing common power rails due to crosstalk issues.
JR. wrote:While I never did one myself I vaguely recall some speed issues associated with swinging the transformer winding with the full audio signal.
I haven't encountered any so called speed issues with floating rails whether it a clipping or driving 10khz signal along with the slewrate exceeding 40V/uS. Already I have a baker clamp installed for each rail to get clean clip recovery. Would love to know more about the so called speed issues if you can show me what i am missing here. The thought of swinging the winding of transformer is not any different than the other way around, only the reference is changed and the commutation of current is always an absolute.

JR. wrote: IIRC QSC did a version with the collectors of a bipolar device amp connected to a grounded heat sink. As I recall there may have been some issues with sensing device current for degeneration or current limiting. QSC had a patent for their current limiting approach that apparently worked, but I never felt it was worth copying.
Yes, QSC is using a grounded collector Class-B PA design, its really a cheap implementation by having an opamp at front stage paired with lower current gain output stage[Driver and output, unlike triple darlington found in Crest Audio amps having bipolar devices] biased at mere 10mA idle current which suffers current starving all the time when loaded with 2 ohm loads gives a horrible clip sticking issues arising from cross-conduction of opposite polarity devices. I don't know which degeneration issue you are pointing out but in QSC they have emitter resistors already in series with every power device like in normal amps and and sensing the current from them is never an issue for those who are skilled in the art. The output averaging short circuit technique which Quilter had invented actually relies on current starving output stage overloading the NE5532 opamp in case of short circuit/over current condition and thereby kicking the opamp to self limit the drive current. It also has a mechanism for boosting opamp +/-15V supplies through the output voltage swing. This trick only works when the output stage has lower current gain[current gain starvation], try this with triple darlington[3 X EF], it will not work effectively because of high current gain available.

My design differs because its the only one of its kind having a vertical Hexfet Mosfets[Not the bipolars] as output devices biased in pure class-A having grounded drain topology and even the biasing transistors which are meant for tracking the idle current are also directly mounted without any insulator on the same heatsink for optimum electrical & thermal performance. The front end is fully discrete with the driver stage configured to give active gate discharging for speed up recovery. Now this might be worthwhile for copying for those interested..... 8-)
JR. wrote: Looks like a nice "winter" amp... I momentarily though about doing a class A variant amp, the last time I came up with a different way but nah.. too busy messing with thermostats and water filters.. and I rarely even listen to music (I got spoiled by having live bands play in my living room, the hifi never sounded the same after that).

JR
Thank you & yes its a good musical heater for winter as they say "Winter is coming" and so is Christmas :D
emrr
Posts: 570
Joined: Sun Mar 02, 2008 10:27 pm
Location: NC, USA
Contact:

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by emrr »

Interesting discussion, thanks for sharing.
Best,

Doug Williams
Electromagnetic Radiation Recorders
User avatar
JR.
Posts: 3700
Joined: Sat May 24, 2008 7:21 pm
Contact:

Re: ||MYWAY|| 50W Class-A with Direct heatsink mounted devic

Post by JR. »

Kanwar wrote:
JR. wrote:The "driven rail" topology is widely done in low cost designs because the front end can be driven from low voltage rails. There can be issues trying to make a two channel amp since they need separate secondary windings.
Yes, it is low cost because the frontend is easy to construct and low cost doesn't always means a cheap design.
While I was at Peavey we actually investigated making a dedicated IC to handle the entire "driven rail" amplifier front end inside a single chip. It was well within the IC process capability voltage and current wise, but the tooling cost was significant, and the junior engineer from the IC company who was assigned to our project was not the sharpest stick in the bunch. We wanted to include a OTA inside the IC to handle our clip limiting and the junior engineer apparently didn't know how OTAs work. When I told him a design constraint was for the OTA to have low input offset voltage in it's LTP to keep control feedthrough low, he suggested adding degeneration resistor to the LTP... duh, degeneration resistors would prevent the OTA from working as intended . :oops: :oops: My lack of confidence in their engineering capability and the cost to tool a proper IC killed the project (probably hundreds of k$) .
Having a floating rails design it requires a separate power supply for each channel, I don't see any problem in that. By the way dual mono amplifiers are always preferred instead of stereo amp sharing common power rails due to crosstalk issues.
I never designed one myself and the transformer issue was second hand information. It may have been specifically about stereo units. Inexpensive power transformers often use bi-filar wires to wind dual identical windings, while the extra coupling may have been inconvenient in a power amp application. I'd be able to give a better answer if I did one from scratch, but I just integrated existing power modules inside larger products.
JR. wrote:While I never did one myself I vaguely recall some speed issues associated with swinging the transformer winding with the full audio signal.
I haven't encountered any so called speed issues with floating rails whether it a clipping or driving 10khz signal along with the slewrate exceeding 40V/uS. Already I have a baker clamp installed for each rail to get clean clip recovery. Would love to know more about the so called speed issues if you can show me what i am missing here. The thought of swinging the winding of transformer is not any different than the other way around, only the reference is changed and the commutation of current is always an absolute.
agreed it shouldn't be an issue since that is a relatively low impedance node all inside the amplifier's feedback loop.
JR. wrote: IIRC QSC did a version with the collectors of a bipolar device amp connected to a grounded heat sink. As I recall there may have been some issues with sensing device current for degeneration or current limiting. QSC had a patent for their current limiting approach that apparently worked, but I never felt it was worth copying.
Yes, QSC is using a grounded collector Class-B PA design, its really a cheap implementation by having an opamp at front stage paired with lower current gain output stage[Driver and output, unlike triple darlington found in Crest Audio amps having bipolar devices] biased at mere 10mA idle current which suffers current starving all the time when loaded with 2 ohm loads gives a horrible clip sticking issues arising from cross-conduction of opposite polarity devices. I don't know which degeneration issue you are pointing out but in QSC they have emitter resistors already in series with every power device like in normal amps and and sensing the current from them is never an issue for those who are skilled in the art. The output averaging short circuit technique which Quilter had invented actually relies on current starving output stage overloading the NE5532 opamp in case of short circuit/over current condition and thereby kicking the opamp to self limit the drive current. It also has a mechanism for boosting opamp +/-15V supplies through the output voltage swing. This trick only works when the output stage has lower current gain[current gain starvation], try this with triple darlington[3 X EF], it will not work effectively because of high current gain available.
Lots of potential veers here... Darlington power devices are harder to bias up for class AB... I did one experimental bias circuit that didn't sense the power devices Vbe and I tested it on a guitar amp using darlington outputs that was notorious for being current starved when cold. Made it hard for music stores to demo in winter months without warming it up first. (That bias scheme never got used because my boss at the time did not have enough balls to try something new in a large scale production design.
My design differs because its the only one of its kind having a vertical Hexfet Mosfets[Not the bipolars] as output devices biased in pure class-A having grounded drain topology and even the biasing transistors which are meant for tracking the idle current are also directly mounted without any insulator on the same heatsink for optimum electrical & thermal performance. The front end is fully discrete with the driver stage configured to give active gate discharging for speed up recovery. Now this might be worthwhile for copying for those interested..... 8-)
For another veer I spent too many bench hours trying to make a conventional class AB amp using vertical MOSFETS several decades ago instead of the more expensive (lower transconductance) lateral MOSFETs from Hitachi). I fried quite a few cheap switching FETs before I gave up (I couldn't get both decent performance and good stability). Another engineer said he was able to make a compound device from a bipolar transistor and vertical mosfet together that worked OK, but I wanted to be able to call it all MOSFET for marketing purposes. After i gave up I turned the project over to a junior engineer to make a MOSFET studio amp using the easier to design with lateral MOSFETs.

Maybe I should have asked you? :lol: :lol: (or cranked in a lot more class A bias to hit a slower part of the transconductance curve) 8-)
JR. wrote: Looks like a nice "winter" amp... I momentarily though about doing a class A variant amp, the last time I came up with a different way but nah.. too busy messing with thermostats and water filters.. and I rarely even listen to music (I got spoiled by having live bands play in my living room, the hifi never sounded the same after that).

JR
Thank you & yes its a good musical heater for winter as they say "Winter is coming" and so is Christmas :D
I scratched out a variant class A based on my biasing scheme but carried to a lot higher class A current in another thread around here somewhere. I was almost tempted to build one, but have other more pressing projects and am out of amp mode these days. I really like the class D approach that Bruno Putzey has pioneered if I was building an amp for personal use. But that would be a "summer" amp project. :D

Merry Christmas.

JR
Cancel the "cancel culture", do not support mob hatred.
Kanwar
Posts: 8
Joined: Fri Nov 29, 2013 8:15 am

||MYWAY|| 50W Class-A with Direct heatsink mounted devices

Post by Kanwar »

JR. wrote: After i gave up I turned the project over to a junior engineer to make a MOSFET studio amp using the easier to design with lateral MOSFETs.

Maybe I should have asked you? :lol: :lol: (or cranked in a lot more class A bias to hit a slower part of the transconductance curve) 8-)
Thanks for the insights JR, Merry Christmas :D

Yes the vertical mosfets are not easy to tame and biasing them in class-A makes them more linear in open loop operation as well, Nelson Pass uses vertical mosfets in his class-A amps for this very reason.

Another familiar approach with vertical mosfets which i have seen:-> It is invented by Paul F. Ierymenko and used in QSC PL6.0 & 9.0 amp, each vertical mosfet is placed in closed loop [feedback taken from source terminal]with a local opamp as a driver there by making it behave like a voltage controlled current source, very linear in operation and no thermal tracking is needed.
JR. wrote: I really like the class D approach that Bruno Putzey has pioneered if I was building an amp for personal use. But that would be a "summer" amp project. :D
Bruno Putzeys has done fantastic work on class-D amplification.
My main field of interest/work is switching power conversion [Class-D & Smps] only, currently working on a self-oscillating delta/sigma 2nd order modulator with output filter LC in a closed loop thus making the output behavior independent of variable load impedance. 8-)
Post Reply