UTHAiM -- Just for Fun

Almost there...

I decided to use my LM1086 regulators, since my trafo could not handle any extra consumption. The trafo is a 50 VA 2x24 V toroid and it's running quite warm with the 32 W load (~10 W losses and ~22 W for UTHAiM boards). Although I did drill the holes ready for my Nazar regs if I decide to upgrade the trafo as well. Would be interesting to test whether I could hear the difference between PSUs.

The front plate still needs some sanding because the logo is a real PITA to remove.
 

Attachments

  • uthaim almost 01.jpg
    uthaim almost 01.jpg
    473.2 KB · Views: 1,048
  • uthaim almost 02.jpg
    uthaim almost 02.jpg
    387.8 KB · Views: 1,035
Last edited:
It's done - finally. I think I might add an Onpow power switch with led ring to the front page but It'll have to wait for now. I'll need to enjoy the sound of this technical marvel first. :D

The amp is hum free, but there's a pink noise type of faint sound when using iPad stock phones, which are low Z and very sensitive. I believe this is the hissing sound that I have heard people talking about with solid state devices. The hiss cannot be heard when I plug in in my LCD-2 phones, which are also very sensitive, so it's not a real problem for me.

The sound seems to be very promising, but unfortunately I haven't had the time for critical listening yet. Oh, and the gain seems to be quite plenty. I'd say ten o'clock on the vol knob is all I ever need. Hopefully I'll find some time to listen more next week.

Edit: About the crossfeed. I have only tried crossfeed with my source device, which is a Raspberry pi with Moode Audio player. Moode has some in-built software crossfeed options and I have tried them all, but quite quickly I have switched them off. It's been a while since I last tested them, but if I remember correctly the bass was not so good and the singers' voice was not so clear. Maybe the hardware versions for crossfeed are more advanced? Anyway thanks for pointing me them.
 

Attachments

  • IMG_0586.JPG
    IMG_0586.JPG
    349.8 KB · Views: 980
  • IMG_0587.JPG
    IMG_0587.JPG
    283.7 KB · Views: 960
  • IMG_0590.JPG
    IMG_0590.JPG
    830.1 KB · Views: 1,004
Last edited:
The 20dB gain is designed for use with Cross Feed.
If you are not using cross feed, you should reduce the gain to say 3x.

In case of the ZGF version, R11 should be reduced to 1.7k.
You can e.g. just solder a 2.49k in parallel to the current 5.1k.
The noise level will also go down by 10dB.


Patrick
 
Class A maximum output current at 180mA bias is >360mA.
At 70 ohm, you are voltage limited (to ~ +/-15V).
Which in turn corresponds to ~ 1.6W for sine waves.

For 16 ohm, you will be current limited at ~1W.
For 600 ohm, ~ 0.18W.


Patrick

Please enlighten me Patrick. Why is the voltage limited to +/-15V when I'm using +/-24V rail voltages?


I think I'll do the modification to reduce gain. I suppose 2,2 kOhm resistor in parallel to R11 would be fine? The resulting R11 = 1,537 kOhm and gain ~10 dB. I have those 2,2 k Vishay resistor in my parts bin.
 
Not all amplifiers in this world is rail to rail. That includes opamps.

The current mirror using MOSFETs will need 9V to work properly.
You can reduce this "wasted" headroom by using BJTs instead of MOSFETs in the current mirror.
You can also reduce the mirror resistor values.
But it still will need ~ 3V headroom per rail.

You can also reduce heat dissipation by using separate (lower) supply rails for the output MOSFETs.
The PCB will accommodate that.
For +/-24V front end, +/-20V would be sufficient for the output stage.
The gain is not so much, but separate supplies is in theory better for the front end.

A lot of these is already in the document and the sited references.


Patrick
 
Today I had some quality time to spend listening to my brand new UTHAiM and here's my two cents trying to describe the sound with a few music examples. I've built up the no feed back version with FETs as current mirrors and Exicon lateral FETs on the output stage. The headphoes I use are Audeze LCD-2.2f, source is RPi with Moode Audio topped with Allo Kali reclocker and Ian's isolator HAT board. DAC is DDDAC connected to RPi with I2S. Music files were flac 16bit/44.1kHz.

First of all everything seems to be really effortless and easy for UTHAiM. Must be due to the quite hefty power reserve it has (1,6 W @ 70 ohm). There's tight and very well controlled bass, control and dynamics hold even when going gets really tough and fully mixed/compressed (Children Of Bodom - Blooddrunk, The Mars Volta - Wax Simulacra).

I also noticed that the soundstage is wider than with my other headphone amps, which is quite impressive for me.

Overal clarity and chrystal clear highs are as well something that must be pointed out (Katie Melua - Shy Boy). Maybe my SE Class A tube amp is a little bit better with the vocals, but very small differences.

Dream Theater's A Change Of Seasons I/II is more engaging and Mike Portnoy's drumming just sounds more fabulous than with my other amps. Hard to describe.

All in all a very good headphone amp and it takes the pole position in my rig. It surely was worth all the effort and I learned a lot building it. Hats off for Patrick for all the help he's given and offcourse huge thanks for publishing his great designs here at DIY Audio.

Now I need to move over to my rig and calm down before sleeping with Kingston Wall - Shine On Me. :hphones:
 
While building the Pioneer Super Linear HPA I had some time to spend with my newly purchased (second hand) Sennheiser HD-800 phones and the UTHAiM. Oh boy what amount of details and soundstage this combination digs from a good regording! Too bad this very good headamp has gotten so little attention around here. I can surely recommend it with both high (HD-800) and low (Audeze LCD-2.2) impedance headphones.
 
Thanks to help from Morde, I curve traced the Renesas 2SK1058/2SJ162, as well as the Exicon ECX10N20 & ECX10P20.
According to datasheets, they should be almost direct replacements.
But apart from Vgs differences, which can vary between FETs, there are also visible differences in Yfs.

Here are some typical results at 500mA bias :
2SK1058 Vgs 1.685 Yfs0.626
2SJ162 Vgs -1.379 Yfs 0.548

ECX10N20 Vgs 0.964 Yfs 0.999
ECX10P20 Vgs -1.423 Yfs 0.885

For the curves,
red is 2SK1058
black is 2SJ162
green is ECX10N20
blue is ECX10P20


Cheers,
Patrick
 

Attachments

  • Renesas Laterals.png
    Renesas Laterals.png
    12.1 KB · Views: 946
  • Exicon Laterals.png
    Exicon Laterals.png
    11.7 KB · Views: 903
  • Like
Reactions: 1 user