Aksa Lender P-MOS Hybrid Aleph (ALPHA) Amplifier

Founder of XSA-Labs
Joined 2012
Paid Member
Following in the footsteps of the Aksa Lender preamp, Hugh and I have been working on a Class A power amp that uses the Aksa Lender front end. We wanted a SE Class A amp with more efficiency than a simple CCS so we decided to try using the very nice Pass Aleph (dynamic bias) CCS. Hugh is using a P-MOSFET as the lower output device, so this is sort of a hybrid, modified Aleph. It is a very simple 6 active amp with no fancy unobtanium JFETs, SITs, hockey pucks, power JFETs, fairy dust Sony FETs, or signal transformers. If you look at the prototype built on veroboard, you will see the usual low cost stuff. The fanciest part I have is a 10uF Silmic II input cap. The rest of the caps are budget Aliexpress. I like to proto with cheap parts, that way there is less pressure in case it flops :) I am using genuine Vishay IRFP's from Mouser though.

Here is the basic circuit and predicted performance specs (it's good for about 20w with 24v rails):
659376d1517198709-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20w-pchan-schematic-png


The circuit uses Hugh's Aksa Lender preamp front end with an Aleph CCS on the top, and a P-channel MOSFET on the bottom driven directly from the Aksa Lender LTP. No CCS on the LTP needed.

Hugh also worked on a monster version of this amp with +/-35v rails and 150w of dissipation per channel to make 53wrms into 8ohm before clipping. To get things rolling, I opted for the +/-24v rail version at a more modest bias of 1.2amps. Slowly work my way up in bias current and then maybe rail voltage (with bigger output devices).

So here is the prototype on a tiny 5cm x 7cm veroboard with two MOSFETs hanging off one end right next to each other. Why so close together? I wanted to use a computer CPU heatsink with heatpipes to cool them - in anticipation of 140w later on. This amp was particularly difficult for me to get running because I made so many wiring errors (a problem when you build it in the wee hours of the morning after a full day of work). Once I found all my errors, the amp fired up no problem. I matched the beta's of the input LTP KSA992's and got perfect 0mV DC offset with no trimpot. You will notice that the tiny (relatively speaking for a Class A) hetsink is fan cooled and has lots of heat pipes (which normally work flipped the other way unless they have capillary wicking inside). But I found them quite effective upside down even as temp was 31C with 55w dissipation. These are salvaged Dell heatsinks for workstation, about $12 to $15. For the record, I am using 1000uF 35v on C3, 22pF NP0 on C4 and C9, and 4x2.2nF (8.8nF) MKT film on C7, 0.47R on R17 and 0.15R on R18. I also have an input RFI filter consisting of 2k2 in series before C1 and a 220pF MKT cap to GND, and 100uF 63v on C2. On C5 and C6 I am using 220uF 100v caps. i had a nice 220uF Silmic II but it was too big to keep on the board and still let me see the resistors unerneath when I was debugging. Otherwise, parts are used as shown on schematic.
659377d1517198709-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-prototype-01-jpg


Closeup of the amp:
659378d1517198709-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-prototype-02-jpg


Without heatsinks (I eventually removed the Silmic II 220uF 50v Aleph feedback cap to make room for debugging):
659473d1517249318-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20w-prototype-bare-jpg


Listened to it for a few tracks and must say, it sounds quite nice and engaging. I know the sound. Foot tapping good - typical Hugh Dean style amp. :D

So here is the first measurement of this simple unassuming amp that takes the best of all worlds. 1kHz 2.83vrms into 8ohms and we get 0.027%THD but in typical Aksa style, mostly in second harmonic, and 10x less in third harmonic, and not much else. This is running well below the recommended 2.0amp bias. So I expect the THD to drop. But the noise floor looks great - actually very good for a linear PSU - perhaps the best I have I seen for one of my linear PSU amps. I am using a basic 100VA donut with 18VAC and a Project16/Prasi CRC PSU with 22mF caps all around and 0.235R on the CRC.
659379d1517198709-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-1-19amp-2-83vrms-8ohms-fft-jpg


I will try boosting the bias current next, and then maybe try to go for the monster 53wrms amp with 35v rails :)

Hope you enjoy it - and if enough interest, JPS64 can design a splendid PCB for us to use. Thanks to Hugh for another gem to the DIYA community. And thank you to Mr Pass for the wonderful Aleph CCS.

Here is the latest Schematic for the Big Boy 52W version - now at a little under 3amps bias and 90W dissipation per side, still need two CPU coolers per channel.

660208d1517584966-aksa-lender-mos-hybrid-aleph-alpha-amplifier-fciehdohbfnkegdm-png


Edit Feb 11, 2018: the Alpha 20 PCB is finally built and tested.
661735d1518324741-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha20-amp-build-02-left-jpg


It measures very well here with 1.31amps bias and produces 0.0069%THD for 2.86vrms into 8ohms:
661737d1518324741-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20-2-87vrms-8ohms-fft-png


ALPHA 20W schematics and stuffing guide for v4 (production boards) here:
Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier - Page 12 - diyAudio

660029d1517503984-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20w_schematic-v4-png


660032d1517503984-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20w_topview-png


660030d1517503984-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20w_3d_render-png


Edit Feb 20, 2018:
BOM v1 for ALPHA 20W PCB layout v4 (above schematic) here:
Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier - Page 51 - diyAudio

Edit Mar 5, 2018: premium components build ALPHA20:
666395d1520249706-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-premium-build-02-jpg


It does indeed eek out a bit lower distortion with Vishay Dale resistors and all Wima and Silmic caps, now about under 0.006%THD for 2.75vrms into 8ohms:
666397d1520249706-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-ch2-fft-png


Edit March 25, 2018 - Alpha 20 for 4ohm loads version:
Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier - Page 124 - diyAudio

Ignore the triple outputs - purely used to simulate a single larger 500W capable IXYS MOSFET pair which will not need to be matched. Hugh does not recommend using triple output IRFP’s

669823d1521533665-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha4r-gif


Edit April 4, 2018: BigBoy is alive and singing! The Beta build went together almost perfectly. Only one minor circuit error on the grounding of the LTP smoothing cap (fixed with a jumper). Amp starts up uneventfully and without any adjustment to achieve 3.0amps bias current and 1mV DC offset.

672472d1522616478-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-test-mono-fast-closeup-jpg


Here is 34wrms into 8ohms from +/-27.6v rails:
672327d1522552973-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-sound-oscope-02-jpg


Here is FFT at 2.84vrms into 8ohms:
672545d1522650414-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-no1-2-95amp-2-84vrms-80hms-fft-jpg


Edit April 4, 2018 - here is the updated as-built schematic for Alpha 20 with 1.35amp bias and no turn-on thump feature. Note that if all pots are removed the amp is a simple adjustment-free Class A amp with DC coupled output that has DC offset within 10mV (typical) and self-tracking, bias current is auto-regulated and very stable:
673097d1522861468-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20-schematic-updates-1-35amps-built-png


Edit April 7, 2018: Alpha B.B. in stereo!
673651d1523088129-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-stereo-build-04-jpg


FFT of Alpha BB at 2.85vrms into 8ohms
673787d1523165499-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-2-854vrms-8ohms-fft-ch1-jpg


Edit April 10, 2018: Summary of various settings to change amplifier configuration bias setpoinysnand speaker impedance driving capabilities. Courtesy of Danny_66: Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier - Page 166 - diyAudio

Edit April 13, 2018: Simulation LTSpice files of Alpha 20 4R and 8R by Danny_66 here:
http://www.diyaudio.com/forums/soli...id-aleph-alpha-amplifier-172.html#post5404051

Edit April 19, 2018: 10kHz 17vpp square wave into 8ohms on Alpha BB:
675635d1523938773-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-bb-5mhz-osc-study-10khz-square-wave-jpg


Edit Aug 23, 2018: several people have contacted me about ramping up for a new build. So here is a recap of suggested schematic for a 4ohm Alpha 20 (which probably puts out closer to 35w into 4ohms):
674314d1523391739-aksa-lender-mos-hybrid-aleph-alpha-amplifier-alpha-20-schematic-updates-2-4amps-4r-png


Edit Sept 4, 2018: Tips for 4R capable circuit...
Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier

Edit June 11, 2019: revised table of source resistors for setting bias current in Alpha 20 for 8ohms and 4ohms by Danny66:
Aksa Lender P-mos Hybrid Aleph (ALPHA) Amplifier
 

Attachments

  • ALPHA-20W-Pchan-Schematic.png
    ALPHA-20W-Pchan-Schematic.png
    287.2 KB · Views: 21,118
  • ALPHA-Prototype-Working-01.jpg
    ALPHA-Prototype-Working-01.jpg
    450.6 KB · Views: 18,346
  • ALPHA-Prototype-Working-02.jpg
    ALPHA-Prototype-Working-02.jpg
    463.3 KB · Views: 20,515
  • Alpha-1.19amp-2.83vrms-8ohms-FFT.jpg
    Alpha-1.19amp-2.83vrms-8ohms-FFT.jpg
    146.1 KB · Views: 18,923
Last edited:
Founder of XSA-Labs
Joined 2012
Paid Member
Definitely interested in the Alpha. How would you say the sound compares to the MoFo, X?

I don't have it in stereo and not connected to my main speaker system so can't really compare. However, I did plug in my headphones in pseudo stereo and it sounds great. Zero background noise with inputs shorted, music is clean, dynamic and very neutral. With the HE400i's it can really make those cans sine with gobs of power. Very articulate bass and clear resounding highs. I would say it sounds a bit drier than MoFo which has maybe almost 8x the H2 distortion. Both have a similar profile though. I am going to wait for JPS64's PCB's this time rather than manually build a stereo second amp.

I think this amp is more neutral in presentation and soundstage and imaging should be just as good as MoFo despite MoFo being zero global feedback.
 
Founder of XSA-Labs
Joined 2012
Paid Member
I switched R17 for a 0.33R metal thin film to boost bias current to 1.62A. Here is the FFT for 2.83vrms into 8ohms, THD is now 0.0096% and mostly H2 (not sure why the PSU is noisier here):
attachment.php


But if you use a metal oxide 0.33R, you get dominant H3 and higher distortion, so not a good idea to use metal oxide resistors for the source resistor:
attachment.php
 

Attachments

  • Alpha-1.6amp-2.83vrms-8ohms-FFT-Thin-Film.jpg
    Alpha-1.6amp-2.83vrms-8ohms-FFT-Thin-Film.jpg
    145 KB · Views: 15,939
  • Alpha-1.6amp-2.83vrms-8ohms-FFT-Carbon.jpg
    Alpha-1.6amp-2.83vrms-8ohms-FFT-Carbon.jpg
    145.2 KB · Views: 15,857
A few words about the ALPHA 20W from the design POV.

There are two significant issues with audio amps. Crossover artefacts, usually H5 and higher odd order harmonics; and harmonic profile, which influences how the ear perceives the sound, that is, natural, or unnatural.

Crossover distortion happens when there are transconductance changes as one device hands over to another, back and fro, across the 0V point on the waveform. It's particularly evident with Class AB amps, because the inactive device is actually switched off, while the active device is going hard at it. Turning on and off semiconductors is difficult because the ear perceives the subliminal high order harmonics which are sprayed into the signal output as 'distortion'. Class A amps fix this problem because neither of the output devices turn on or off; they may current cycle up and down, but they do NOT turn off, and so the death throes 'artefacts' are not created so the sound is more 'natural', whatever that is........ The ALPHA output devices do not switch, although they do current/voltage rise and fall with the load.

Lastly, the harmonic profile has been thrown into prominence in the last ten years, but in fact has been pivotal for good sound and Jean Hiraga demonstrated this in Le Revue du Son in Paris in the mid-sixties. He said that if you can arrange the first five or six harmonics off the fundamental along a straight, descending line, with at least 10dB reduction with successive higher harmonic, the sound will be perceived as 'natural'. H2 should start no higher than 68dB down from the fundamental. This was a controversial statement at the time, and even now, because it pays little regard for THD, which is a rms summation of the harmonic percentages of the fundamental, with no information about each actual harmonic. I have been designing amps using this philosophy for twenty years, and if you give regard to the monotonic decrease of the harmonics and a moderate THD (no more than 0.03%) you can create a good sounding amplifier. There are other issues, of course, but it certainly explains why many amps with very low THD still do not sound good. The fact that this is a subjective assessment, and humans enjoy squabbling about the subjectives has only clouded high end. During the eighties the Japanese opened up the 'THD Wars', to very little benefit. Still a lot of people like single ended triode amplifiers, and they deliver outrageous THD figures.

I do not go for Class A amplifier as a rule. They are brute force machines, with less efficiency than a gasoline engine, usually less than 20%. Even on this 20W amp with 24V rails, the output dissipates around 96W, and for only 19W this is 19.8% efficient with a lot of heat. The real skill is in designing good sounding AB amps, and all my commercial amps are AB with very high efficiency and perceptible natural sound with deep stereo sound field. That said, the ALPHA uses Nelson Pass sliding CCS in the output stage and used here with little change; it's a brilliant, creative solution to increase the efficiency of a simple Class A in SE, where the output currents cycle up and down but nothing turns on or off.

In sound field for this amp, I'd suggest building it, but we assured that all my commercial amplifiers are just a little better and much more powerful. But you'd have to pay for them - the ALPHA is cheap as chips, excepting of course the heatsinking......

My sincere thanks to XRK, a fantastic guy to work with. He should take great care his fingers never melt, he works so fast.......

HD
 
Last edited:
  • Like
Reactions: 1 user
Founder of XSA-Labs
Joined 2012
Paid Member
Hi Dadod,
Ah, the 29 transistor amp! You can see I have no problem making 5 or 6 transistor jobs :). The hurdle on your GW3 amp is requirement to match a bunch of BC550/560’s. I don’t want to build it without the matching. So the extra steps get in the way sometimes. I want to do another headphone comparison thread though and would like to include your amp so hope to build it in next month.
Cheers,
X
 
Founder of XSA-Labs
Joined 2012
Paid Member
Just an update that the Master layout artist JPS64 is on the case. He will make two layouts: the 20W board and the 55W board. I think the 20W made to allow compatibility with the DIYA Store’s UMS specification might be a good idea? Whilst the 55W is strictly CPU cooler territory. I don’t think anyone wants a passive heatsink that can handle 300w per channel. :D

So outputs will be three per side closely spaced to all fit on copper block of the Dell surplus workstation heatsinks good for 130w to maybe 150w per side. The Dell FD841 heatsink designed for Xeon 5500 cpu’s Is supposed to be good for 130w with heatsink at 41C. We can tolerate 55C heatsink temps so perhaps can push 150w or slower fans speeds.

Just search eBay for Dell FD841 T7400 heatsink. About $17ea.

I am not choosing new COTS CPU coolers because modern CPUs are much smaller and there isn’t enough area to fit 3 TO-247 packages side by side.

Look how big the copper pad on the FD841 is:
714MX-8lATL._SL1500_.jpg


If someone can find a modern heatsink this big with heatpipes and a fan on Amazon, please let us know.
 
Last edited:
  • Like
Reactions: 1 user