"Millett Hybrid" redux: NuHybrid headphone amp, using the Korg Nutube
Mar 18, 2017 at 2:35 PM Post #31 of 507
OK.  With the help of a very good friend - Dsavitsk of ECP Audio - we've solved the oscillation issue with the volume pot.  There is now no noise whatsoever in my volume pot travel.  Gate stoppers on the MOSFETs fixed it.  A Gate stopper consists of a 1K resistor inserted between the Gate leg of the MOSFET (center pin) and the center pin connection on the PCB.  It takes some solder acrobatics, but most of you will probably have an easier time of it than I did, since I had already trimmed the MOSFET pins pretty short when I first built the NuHybrid.  Here's some pics -
 
This is Q2 and Q3, with the 1K resistor inserted between the Gate leg of each MOSFET and the other lead of the resistor inserted into the Gate leg's pad on the PCB:

 
Basically, what you are looking at is one resistor lead is bent down vertically and inserted into the center hole on the PCB for the MOSFET.  The other resistor lead is bent 180 degrees back over toward the MOSFET, whose Gate leg (center pin) is bent back up horizontally and soldered to the resistor lead on top of the resistor.  As I said - yours probably won't look like this if you still have long leads on the MOSFETs.  Note that I had to relocate the R19, R20, R22, and R23 resistors to the bottom of the PCB to make room for the Gate stopper resistors:
 


You can also see the 0.33uf Sonicap Gen IIs that I used for bypasses on the output caps.
 
Same Gate stopper applied to the MOSFET in the relay-delay:
 


This MOSFET is for the relay.  It's not really in the signal path, but Dsavitsk suggested there might be some bleed-over if it oscillated, too.  So, I put one here, too - just in case.  The Gate stopper resistor had to twist over to the side because of C1.  Be very careful if you do it this way that you don't short one of the outside legs of the MOSFET with the Gate leg and resistor lead.
 
Also note the C2 cap was upsized.  This is a very inexpensive Murata 47uf 25V electrolytic.  25V is probably too low a rating, but I'm using a linear-regulated walwart at 24VDC anyway, so I doubt seriously it's going to be exposed to anything more than 24V.  Still, the point is that you can find a bigger cap pretty easily (no high quality needed here) that should fit without any trouble.  In my case, this extended the delay on power-up from about 6 seconds with the original cap to something over 30 seconds.  Yeah, that's a hassle to wait that long, but when it kicks in now, there's only a very low "tap" that you can hear.  Although, the "tap" sounds like a spoon tapping on a crystal glass.
wink.gif
  Before, the relay was engaging before my output caps had fully charged and was causing a nasty voltage spike and thump.  There is still a turn-off thump, but I'm used to dealing with that a bit more than a turn-on one.
 
Here are a couple of overall pics -

 
and catching the glow from the NuTube -

 
Despite the camera setting, the NuTube glow is easily seen in daylight, unless you have some glare on the transparent NuTube cover.  You can compare the green with the LED near the power switch.  The NuTube glows with a very nice "ocean green" color just like some LEDs that I've sold as options on the Starving Student.
wink.gif

 
One other thing to note - the NuTube definitely has some microphonics, but it takes a real wallop to hear it.  Unlike glass tubes with microphonics, the NuTube will "sing" at a very high-frequency pitch - like the crystal tap mentioned above.  I will say that the only thing that causes this is being very rough and forceful when plugging in your headphones.  There are a couple of reasons I was doing this - 1) I was plugging in after powering up, until I was certain that the turn-on thump was gone, and 2) with that little PCB setup on some metal standoffs, it takes some forceful action to hold it and plug the headphones in.  If you are gentle and insert the jack slowly, you will not start the NuTube ringing.  A good thing to mention is that nothing else I've done will cause it to ring - tapping it directly, shaking the PCB, or tapping anything else on the PCB.  It only occured when violently shaking everything up plugging in the phones.
 
So how does it sound?  I've been doing a lot of FM tuner work lately, so that's all I've heard it on thus far.  However, I directly compared it to a pair of Mini3s and PIMETAs that I've been using in that work.  The NuHybrid compares favorably to the Mini3, at least in this setup with a highly-modded Sony ES tuner.  Detail is good, bass slam is great (note that I used 1000uf on the output caps) and it sounds very dynamic.  The PIMETAs are smoother, but I think that may be a combination of the OPA627s in the PIMETAs, along with a very low gain.  That's the last issue, I think - the NuHybrid has very high gain.  I don't see any feedback loops except on the opamps and those are set to unity.  Pete - if you have any suggestions on how to tame the gain, it will be greatly appreciated.
 
I'll report back when I get in some listening time on a quality DAC.  Thank you Pete for this new toy and thanks to Dsavitsk for the oscillation solution!
 
Mar 18, 2017 at 4:04 PM Post #32 of 507
Thanks Tomb.

I think an even simpler solution will be to simply replace the 2N7000 FETs in the buffers with 2N3904 NPN transistors. I haven't tried it yet, but it should work just fine, and may have lower distortion than the MOSFETs.

I'll try it when I get home Sunday.

Pete
 
Mar 18, 2017 at 6:58 PM Post #33 of 507
Thanks Tomb.

I think an even simpler solution will be to simply replace the 2N7000 FETs in the buffers with 2N3904 NPN transistors. I haven't tried it yet, but it should work just fine, and may have lower distortion than the MOSFETs.

I'll try it when I get home Sunday.

Pete

Cool.  Yes, that would be a good deal simpler than what I did with the MOSFETs.
wink.gif

 
Any suggestions on lowering the gain?
 
Mar 19, 2017 at 2:53 PM Post #36 of 507
The Kemet relay is out of stock in the Swedish Mouser (not sure if they use the same stock as other countries or not), does anyone know of a good replacement that could work intead?
Also, the 2N7000 FETs are oos, but I'll wait and order until Pete have tested the suggested 2N3904 NPN transistors instead.
 
Mar 19, 2017 at 6:57 PM Post #38 of 507
Hfe on those is max 300. At the 10mA running through these, the base current will likely drag the voltage too much. A darlington (2N6426 perhaps?) might work.


Yes, we'll need to lower the current through the transistor. I'll try 10k instead of 1k on the emitter and see if it measures OK.

Pete
 
Mar 20, 2017 at 8:28 AM Post #39 of 507
The Kemet relay is out of stock in the Swedish Mouser (not sure if they use the same stock as other countries or not), does anyone know of a good replacement that could work intead?
Also, the 2N7000 FETs are oos, but I'll wait and order until Pete have tested the suggested 2N3904 NPN transistors instead.

...and apparently I didn't even read the first page of this thread - I see now that this question have already been answered there 
blink.gif
 
 
Btw, I think I spotted another small mixup between the BOM and the actual schematics - C6 & C7.
 
In the BOM C7 is a 470uF Nichicon electrolytic, but it should really be C6.
Likewise, C6 in the BOM is one of the WIMA film caps, but it should be C7.
 
I guess it should be obvious to everyone trying to assemble the amp, but I thought it could be worth mentioning anyway.
 
Mar 20, 2017 at 11:39 AM Post #40 of 507
  Hi Pete, since this amp use both the nutube for gain and opa551 for output buffer like the sankanagu, do they share any similarity in sound signature?

 
There is some similarity.  If anything this one is a bot more "tubey".  On the Sangaku I use a much higher B+ voltage, and also a tiny bit of NFB around the tube.  That way it can drive a much higher output signal without too much distortion.
 
Pete
 
Mar 20, 2017 at 11:40 AM Post #41 of 507
 
In the BOM C7 is a 470uF Nichicon electrolytic, but it should really be C6.
Likewise, C6 in the BOM is one of the WIMA film caps, but it should be C7.
 
I guess it should be obvious to everyone trying to assemble the amp, but I thought it could be worth mentioning anyway.

 
Thanks!  I will fix the documentation.
 
Pete
 
Mar 20, 2017 at 12:47 PM Post #42 of 507
OK...
 
I've made an update to the design to eliminate instability in the 2N7000 buffers when the volume control is near zero, and to reduce the power-on noise in the headphones.
 
Q2 & Q3 are changed to 2N3904 transistors.  R3 & R4 are now 1k, and R21 & R24 are 10k.  The resistor quantities in the BOM remain unchanged, basically two pairs were just swapped around.  Q1 remains 2N7000 FET.  So, the only new components are Q2 and Q3.
 
I measure no differences between using the 2N3904 and 2N7000.
 
I have updated the documentation and the Mouser BOM.
 
For you early adopters in the USA (postage outside the USA is 10 times the cost of the transistors) that already bought parts, send me a PM and I'll drop a couple of 2N3904's in the mail for you.  Sorry for the trouble!
 
Pete
 
Mar 22, 2017 at 8:14 PM Post #43 of 507
  OK...
 
I've made an update to the design to eliminate instability in the 2N7000 buffers when the volume control is near zero, and to reduce the power-on noise in the headphones.
 
Q2 & Q3 are changed to 2N3904 transistors.  R3 & R4 are now 1k, and R21 & R24 are 10k.  The resistor quantities in the BOM remain unchanged, basically two pairs were just swapped around.  Q1 remains 2N7000 FET.  So, the only new components are Q2 and Q3.
 
I measure no differences between using the 2N3904 and 2N7000.
 
I have updated the documentation and the Mouser BOM.
 
For you early adopters in the USA (postage outside the USA is 10 times the cost of the transistors) that already bought parts, send me a PM and I'll drop a couple of 2N3904's in the mail for you.  Sorry for the trouble!
 
Pete

Pete, I was about to sit down and populate the PCB when I saw this. I don't have any 2N3904 (I'll be emailing you shortly). If I wanted to build the NuHybrid now and swap out Q2 and Q3 when I get the right transistors, should I still swap those resistor pairs?
 
Mar 22, 2017 at 9:32 PM Post #44 of 507
  Pete, I was about to sit down and populate the PCB when I saw this. I don't have any 2N3904 (I'll be emailing you shortly). If I wanted to build the NuHybrid now and swap out Q2 and Q3 when I get the right transistors, should I still swap those resistor pairs?

 
Yes, use the updated resistor values, even with the 2N7000.  I just tried it, and actually the oscillation goes away with 10k resistors on the followers.  So maybe no need to change to 2N3904 at all.
 
I'll do some more measurements tomorrow to confirm that everything is OK.
 
Pete
 
Mar 23, 2017 at 1:45 AM Post #45 of 507

 
Complete, but I experience a HF ringing/oscillation noise on power on or whenever the NuTube sees any vibration (it seems crazy microphonic) that slowly fades away over time. Sounds kind of like tapping on a wine glass, but the ringing takes an impossibly long time to fade away. Comes back if I tap the NuTube or even touch the amp in any way. Tube noise also seems to be abnormally high (very audible on Fostex TH-X00). I swapped the resistor values but I used the 2N7000 transistors in a the Q positions. It does amplify music though. Where should I start debugging this?
 

Users who are viewing this thread

Back
Top