There is also Hugh’s own headphone amplifier design, the Silicon Harmony. A 5w into 8ohm SE Class A amp with zero global feedback. It has a harmonic profile similar to SET tube amp but less THD so as not to go overboard with the “Tubey” sound.
It’s the best headphone amp I have heard - although not a public DIY design. Achieves about -66dB for H2 and -90dB for H3 and nothing else. Can drive any headphone load.
Can you to describe about their sound ?
If you describe one of Hugh’s amps, you basically can describe them all as they share similar sound quality goals:
Perhaps natural sounding with a great sense of dynamics and engagement (makes you want to tap your feet type of musicality). They are neutral in frequency response so not clothed or anything of course (measures flat). This one being a zero global feedback design, the soundstage is very open and precise. Absolutely black noise floor when source is turned off or disconnected. My Pocket SE Class A amp sounds similar (same BF862 front end) but Silicon Harmony has a true CCS and a KSC1845 based CFP using a P-channel MOSFET output stage.
Perhaps natural sounding with a great sense of dynamics and engagement (makes you want to tap your feet type of musicality). They are neutral in frequency response so not clothed or anything of course (measures flat). This one being a zero global feedback design, the soundstage is very open and precise. Absolutely black noise floor when source is turned off or disconnected. My Pocket SE Class A amp sounds similar (same BF862 front end) but Silicon Harmony has a true CCS and a KSC1845 based CFP using a P-channel MOSFET output stage.
Hi Steve,
Perfectly normal, caps charging. It's infrasonic and will not damage the speaker.
In modern cars when you turn on the ignition the revs turn quickly to 8000rpm then fall back to zero. When you start the engine the rev counter is totally accurate. This excursion of the speaker cone is analogous to this automotive quirk.
HD
Perfectly normal, caps charging. It's infrasonic and will not damage the speaker.
In modern cars when you turn on the ignition the revs turn quickly to 8000rpm then fall back to zero. When you start the engine the rev counter is totally accurate. This excursion of the speaker cone is analogous to this automotive quirk.
HD
If the LTP smoothing cap was a small 47uF instead of the 1000uF, there would be a faster charge up and a thump sound.
Hi Danny,
You are limited with compensation cap calculation in Spice. You are dead right, this spatial aspect is not measured in audio and therefore conventional thinking is that it is not important, the arrogance of conservatism! People say Spice can calculate compensation caps; I have found that listening tests are more definitive, though the simulation will indicate something within +/-30%. You can do a Bode graph of course, but the Spice engine does not take account of the parasitics in the board, and certainly does not fully describe the active devices, or even the HF performance of the electros. Ergo - take an educated guess with Spice, then fiddle up and down, listening careful for spatial cues, particularly with a large orchestra well recorded.
When you increase the Closed Loop Gain, you decrease the loop gain, increasing higher levels of low order and decreasing high order. Correspondingly, if you decrease CLG you increase loop gain, that is, increase the levels of global feedback, and this creates more error correction of higher orders, leading to more distortion of H5 and beyond. I try to get H5 below -100dB, and so clearly here you have a balancing act, like just about any other machine in the physical world.
You could try 22pF, 30pF and 36pF and have a listen! I cannot tell you what is ideal, but clearly this cap will be dominated not just about the loop gain but also by the target impedance, that is, the shunt fb resistor. 2k2 will need more cap than 2k7, of course.
HD
Hugh:
I've been thinking about what you said about the important impact of the compensation capacitor on the spatial aspect: I must admit that I was convinced that the spatial image depended on other factors, too. It is not easy to figure it that way and it is not easy to find a generic rule to apply it to other circuits.
That concept that you have described it does not make me easy to extend its application to other schemes (with different open-loop gains, different closed-loop gains, poles located at other frequencies, etc.).
Although, also, you would have to admit that it is not easy for us to accept that we must experiment to find the optimal value to find that magic point in the spatial aspect. I ask, with all the respect you deserve, that given the technical training I have, I have a hard time accepting that this concept can not be associated with a rule of general and conducive application. I am aware that circuit simulators diverge in the high frequency analysis with respect to the real situation, although, with some astusia and knowledge, additional factors could be incorporated to bring both results closer together.
Best regards
Last edited:
X,
I have the 1000uf cap in place of 47uf C103.
Also no trimmers at R125 R1270 as per updates.
1050ohm resistor for R105.
It's strange to watch the speaker cone slowly fully retract then relax back.
Steve.
I have the 1000uf cap in place of 47uf C103.
Also no trimmers at R125 R1270 as per updates.
1050ohm resistor for R105.
It's strange to watch the speaker cone slowly fully retract then relax back.
Steve.
X,
I have the 1000uf cap in place of 47uf C103.
Also no trimmers at R125 R1270 as per updates.
1050ohm resistor for R105.
It's strange to watch the speaker cone slowly fully retract then relax back.
Steve.
And it does it movements in slow motion over 3 seconds right? That's exactly what is supposed to happen vs a quick thump sound. So without any trimmers, what is your DC offset when first turned on (after say 5 seconds) and when fully warmed up? Is the bias current rock steady also? Isn't it neat to see an adjustment free amp?
Last edited:
Thank you Danny your figures are all correct.
That's good, I paid attention 🙂
Danny
X Yes around 3 seconds without a sound.
DC offset when first on was 8mv on both channels,So have not checked when fully warmed,I know i'm slack!!!! nor have i checked the bias But i will endeavour to do so if i can just stop listening to the amp.
And Yes it could not be any easier to build without any adjustments although if anyone had seen me trying to match the 4 KSA 992 transistors.
Steve.
DC offset when first on was 8mv on both channels,So have not checked when fully warmed,I know i'm slack!!!! nor have i checked the bias But i will endeavour to do so if i can just stop listening to the amp.
And Yes it could not be any easier to build without any adjustments although if anyone had seen me trying to match the 4 KSA 992 transistors.
Steve.
Hi Diego,
Always a pleasure to chat with another Spanish speaker who writes better English than most English speakers on this forum!! You, Maty Tinman, and many of the European and sub-continent guys are just amazing. I have less ability with other languages than a garden gnome, hats off to you all!
I am uneasy about this method of finding the compensation capacitor too. It sits badly with the usual scientific method in every way. I put it to you that this is the reason this engineering process is also art. There is judgement involved, like designing the fb for an electric power steering system for an auto. It is empiricism, not the scientific, mathematic approach.
It is almost impossible to be objective about the subjective. That is the problem......
Try it. You will hear the changes. BTW, I have used the Rush cascade for many years, it has very low phase shift and blinding speed. I like your circuit, Diego, it's masterly.
Ciao,
Hugh
Always a pleasure to chat with another Spanish speaker who writes better English than most English speakers on this forum!! You, Maty Tinman, and many of the European and sub-continent guys are just amazing. I have less ability with other languages than a garden gnome, hats off to you all!
I am uneasy about this method of finding the compensation capacitor too. It sits badly with the usual scientific method in every way. I put it to you that this is the reason this engineering process is also art. There is judgement involved, like designing the fb for an electric power steering system for an auto. It is empiricism, not the scientific, mathematic approach.
It is almost impossible to be objective about the subjective. That is the problem......
Try it. You will hear the changes. BTW, I have used the Rush cascade for many years, it has very low phase shift and blinding speed. I like your circuit, Diego, it's masterly.
Ciao,
Hugh
Question please, I have used FQA36P15 and FDA38N30 mosfets as suggested by Hugh right at the start of the thread. I need to get some more but my usual wholesaler is out of stock of the FDA38N30. Can you tell me is FQA40N25 a good substitution as those are in stock.
I am not very familiar with Fets.
Regards
Alan
I am not very familiar with Fets.
Regards
Alan
Yes, Alan, similar device and there will be almost no sonic difference.
I have used both with equal outstanding results, but the FQA has much more transconductance, 29S versus 6S, and this will slightly raise the upper clip point about half a volt. Ciss is 3.5nF compared to 2.6nF and very similar current and power ratings. No real differences sonically.
HD
I have used both with equal outstanding results, but the FQA has much more transconductance, 29S versus 6S, and this will slightly raise the upper clip point about half a volt. Ciss is 3.5nF compared to 2.6nF and very similar current and power ratings. No real differences sonically.
HD
Last edited:
The fb compensation cap is C113, the one parallel to the 22k R113 resistor.
You can tune that cap for best imaging stage, +-30% of value
see previous posts from Hugh, X and Bigun.
Grts,
Danny
You can tune that cap for best imaging stage, +-30% of value
see previous posts from Hugh, X and Bigun.
Grts,
Danny
Last edited:
You, Maty Tinman, and many of the European and sub-continent guys are just amazing
Hugh, if one day you travel to Tarragona I invite you to a very good seafood lunch and I will show you and comment on the Roman remains -the best in Europe after Rome. And the Cathedral -early gothic style.
Tarraco - Wikipedia
- Home
- Amplifiers
- Solid State
- Aksa Lender P-MOS Hybrid Aleph (ALPHA) Amplifier