Aksa Lender P-MOS Hybrid Aleph (ALPHA) Amplifier

Jump to Latest
xrk971

xrk971

Founder of XSA-Labs
Joined 2012
etsy.com/shop/XRKAudio
Paid Member
Also, the feedback connection points are different. The ground lift was just extra insurance against hum from ground loops. There is an extra diode to protect the input stage as well. Mostly though, the PCB was designed with a wide variety of interface bolt holes for all the popular and low-cost Intel/AMD CPU coolers that use heatpipes/radiator/fans.

Whereas the Alpha 20 was designed to be UMS compatible.
 
Last edited:
xrk971

xrk971

Founder of XSA-Labs
Joined 2012
etsy.com/shop/XRKAudio
Paid Member
xrk971 said:
Ok, so I did the test with my Alpha 20 running 1.35amp bias with R131=0.47R and R132= 0.22R.

I used two 270ohm (5W) resistors in series for the high impedance case for 3.52vrms or 9.96vpp.

Then with a 10.0ohm 2% 10W resistor I measured 3.39vrms or 9.59vpp.

So (1-9.59/9.96)=0.0368 or 10ohm/0.0368=27.2 DF - fairly close to the 30 DF Hugh predicted for the Alpha 20 running 1.93amps bias.

So one thing that is important is to use very low impedance speaker cables. They do make a difference. 16ga stranded copper zipcord is not as good as solid copper 12ga Romex. So back goes the Romex into the speaker setup.
Click to expand...

I forgot that Hugh designed the Alpha to have higher damping factor at low frequencies where you need the bass cone control. Less at higher frequencies where you want less global feedback for better imaging. I ran my test at 1kHz. Maybe I should try redoing it at 55Hz kick drum center frequency and see what t is?
 
I

irribeo

Member
Joined 2013
xrk971 said:
I forgot that Hugh designed the Alpha to have higher damping factor at low frequencies where you need the bass cone control. Less at higher frequencies where you want less global feedback for better imaging. I ran my test at 1kHz. Maybe I should try redoing it at 55Hz kick drum center frequency and see what t is?
Click to expand...
And maybe take 8 ohm load, Danny's sym shows the higher low frequency damping seems to disappear a little too ???
 
danny_66

danny_66

Member
Joined 2005
I just read an interesting post about damping factor.
See the post of atmasphere.
I copied a part of it:

Amps that exhibit 'tight' bass often have lots of feedback, which is there to reduce distortion.
Unfortunately a price is paid: loop feedback at low frequencies can behave as a sort of dynamic compression,
and has the ability to cause the amplifier to loose soundstage definition (the 'air' around instruments that flesh out the body of the instrument as well as its ambient signature).
Often 'tight' bass is as a lack of definition at low frequencies.

IME, when you start to loose definition, the of the first things to go is the low frequency ambiance in the recording.
Further reduction of definition results in 'poster board' images of instruments in the soundstage.
Initially, this might sound as if the amp is more focused, but after a while the lack of ambient signatures (hall reflections and the like) let you know what is really going on.


See also the post from folkfreak:
As the owner of a pair of VTL MB 450 III with adjustable damping factor (4 settings) I can attest to the fact that damping factor matters.
Obviously it's an interaction with the rest of your system but as I got my room resonances under control I could lower the DF on my Magico Q3s and now have it at the lowest setting.
The effect of lowering DF is to lose some immediate "punch" in the bass but replace it with greater air, scale and overall body which is much more preferable.
My suspicion would be that too many amps (especially solid state) are over damped but that's just my preference and I'm surprised more manufacturers don't offer adjustable DF
 
Last edited:
danny_66

danny_66

Member
Joined 2005
Just an idea:
Now there are 2 options for the Aleph damping factor, depending on where the fb is taken: before or after the sensor resistor.
If you "split" the sensor resistor in 2 or more resistors in series, you get additional DF options or you can customize the DF to the value that you want.

I did some simulations and this works, I can customize the DF :)
 
Last edited:
xrk971

xrk971

Founder of XSA-Labs
Joined 2012
etsy.com/shop/XRKAudio
Paid Member
The dissipation would be the same - the smaller TO-247 packages simply cannot handle more than 35w each. They would melt at 100w. The bias current is pretty much function of the R131 (on Alpha 20 board) source resistor on the CCS. The big TO-264 packages handle more dissipation but have about three times the capacitance. So there is a difference there but Hugh’s circuit takes all of that into consideration and has been optimized (recall it was modeled as three pairs of IRF240/9240 for good reason).
 
Last edited:
danny_66

danny_66

Member
Joined 2005
AKSA said:
Yes, Danny, use a 100R pot across the 0.12R output resistor and take the fb from the wiper, no change to the 22k fb resistor.
Click to expand...

Yes, good idea, that will be the best and easiest way to try the different damping factors.
If I settle for one I can change the pot to two resistors (100r total), values determined by the pot.

This opens back the option for the 4R to go for 0.18 sensor and 0.33 source resistors to have less heat dissipation.
I can still adapt the damping factor if it gets too low.

X, a proposal for future PCBs: place for two extra resistors in series, parallel to the sensor resistor.
At the junction you can take the fb.

Regards,
Danny
 
Last edited:
AKSA

AKSA

Member
Joined 2001
aksaonline.com
Danny,

On the 4R you can reduce the output sensor from 0.18 to the original 0.12 and restore the standard R17, which is 865R. Increasing it over 0.12 whilst still working well at the fixed figure does not give any advantage; it just increase Zout needlessly. You don't want DF less than about 27. Many SET tube amps have a Zout of around 4R, and their bass is less than optimal, so I feel high Zout is not a good idea.

Cheers,

HD
 
I

irribeo

Member
Joined 2013
xrk971 said:
The dissipation would be the same - the smaller TO-247 packages simply cannot handle more than 35w each. They would melt at 100w. The bias current is pretty much function of the R131 (on Alpha 20 board) source resistor on the CCS. The big TO-264 packages handle more dissipation but have about three times the capacitance. So there is a difference there but Hugh’s circuit takes all of that into consideration and has been optimized (recall it was modeled as three pairs of IRF240/9240 for good reason).
Click to expand...

The IRFP240/9240 can't dissipate much more. But 0.33R on Alpha is around 1.9A, 0.27R in 4R version 2.4A, 0.22 would be around 3A, and then at 24V.

Now we have 36V, with around 3A (0.22R) that would be 100W+ ??? WIth 0.11R as in schematic that would be around 6A and 200W+??? So where do I go wrong ?
 
danny_66

danny_66

Member
Joined 2005
Hi Hugh,
Yes, the optimal values for 4R are 0.12, 0.33, 865
I did some simulations near clip level (16.7Vout) and only these values still gave a nice FFT.
Before I did simulations with only 9Vout and those FFTs also looked good with other values, but not anymore at 16.7Vout
Regards,
Danny
 
JPS64

JPS64

Member
Joined 2011
Here the results of my long reflection about I2C VU-METER:
- THAT 4301 dynamic processor and adjusted AN119 (needs +/-15V signal ground referenced)
- low cost 8bit I2C ADC (needs +5V analog and +5V digital signal ground referenced) - ADC121 is 12bit, SOT23-6, only example
- ADuM1250 I2C isolator (needs +5V digital signal ground referenced and +5V digital ground referenced)
- to avoid ground problems by isolating dirty digital ground.
- I2C connector supplied today on ALPHA BB boards
- replacing OLED 1.3 inch with 4.2 inch three colour e-paper display

This need a complete redesign of the actual solution!

IMHO, this could be a standalone developpement (I2C today supplied on ALPHA BB boards) then not only for ALPHA amps!

JP

THAT Corporation 4301 Analog Engine - Dynamics Processor IC
http://www.thatcorp.com/datashts/dn119.pdf
4.2inch e-Paper Module (B - Waveshare Wiki)
YouTube
 
You must log in or register to reply here.
Top Bottom