Go Back   Home > Forums > >

Headphone Systems Everything to do with Headphones

Designing my headphone amp
Designing my headphone amp
Please consider donating to help us continue to serve you.

Ads on/off / Custom Title / More PMs / More album space / Advanced printing & mass image saving
Reply
 
Thread Tools
Old 12th September 2021, 10:29 AM   #171
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
I've tuned this a little. Using a 2M and switching from a 14.5uF to a 0.022uF ! flattens the response at the load considerably

Lines here are for 32, 55, 64, 120, 300 and 600 ohm loads. The 600ohm is the top line near 9.7dB.

Screenshot 2021-09-12 at 11.28.09.png

For kicks and giggles I've added both 4 and 8 ohm loads into the list:

Screenshot 2021-09-12 at 11.33.43.png

I suspect the poor old BC547Bs would throw a tantrum attempting that. -6dB into 8ohms.. you'd hear it quietly!
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 12th September 2021 at 10:36 AM.
  Reply With Quote
Old 13th September 2021, 06:39 AM   #172
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
So starting to work out the final tuning.

The heavy third harmonic is simply this (and a common issue with cascodes) - the top cascode is sitting too close to the top the B+ and on amplification, so the peak of the waveform is clipped.

Screenshot 2021-09-13 at 07.34.51.png

Screenshot 2021-09-13 at 07.34.47.png

I made an adjustment to tune the Vpp down but I may simply use a voltage divider instead. There's not a lot of current at this stage.

There's also a more important secondary issue I need to sort out:

Screenshot 2021-09-13 at 07.43.35.png

Blue - output voltage
Red - input into the NPN in the current mirror
Green - tube output voltage

The resulting blue wave form is misshapen due to the phase shift of the two input waves - both sets of waves go through caps to isolate. The green goes through a large bank of caps to isolate the tube and headphone. The blue is the after the isolation cap of the current mirror. It's not the current but the voltage.

So essentially the caps are not shifting the phase by the same degree. So I will need to work out how to correct this or the amp will be third harmonic heavy.
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 13th September 2021 at 06:48 AM.
  Reply With Quote
Old 13th September 2021, 06:48 AM   #173
UltimateX86 is offline UltimateX86  France
diyAudio Member
 
Join Date: Jun 2004
Location: France
Hi nick, I really like the name of your amp "c'est compliqué"

I have a hybrid to show you

simulation of a hybrid amp : so long
__________________
I do not sell anything, I'm just trying to build a www.circlotron.audio
  Reply With Quote
Old 13th September 2021, 07:04 AM   #174
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
Quote:
Originally Posted by UltimateX86 View Post
Hi nick, I really like the name of your amp "c'est compliqué"

I have a hybrid to show you

simulation of a hybrid amp : so long
Nice! (The wife is French)

I see the link has a final mosfet output stage. I think I made it doubly difficult by attempting to keep the tube output then augment it with a current drive whilst isolating two differing power loops.
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)
  Reply With Quote
Old 13th September 2021, 07:19 AM   #175
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
Quote:
Originally Posted by NickKUK View Post
There's also a more important secondary issue I need to sort out:

So essentially the caps are not shifting the phase by the same degree. So I will need to work out how to correct this or the amp will be third harmonic heavy.
A quick model on the smaller model with a duplicate set of caps has corrected the issue.

Only issue here is that each mirror 'cap' is now a cap bank of 4x330uF and a 0.22uF to mirror the phase shift. For a tube OTL this amount of output capacitance prevents rolloff and allows current (the paralleled caps means less ESR too). In fact in the previous totem PP design, each opposing pair of tubes had a cap bank.

In the hybrid there's no getting away from this either as the mirror isolation should not roll off any low frequencies.

Let me make the modification and let the big sim run. Hmm with paralleled tubes comes the fun of varying parallel capacitance which then could cause a variable phase shift like a varistor.

First wave (hence misshaped) of the model shows they're in phase:
Screenshot 2021-09-13 at 08.37.21.png

Will allow the model to run for 10 sim seconds..
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 13th September 2021 at 07:38 AM.
  Reply With Quote
Old 14th September 2021, 05:33 AM   #176
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
So I spent yesterday evening going back over the cascode.

I'm aware that LTSpice gives a rough simulation vs real results however as DF96 used to say it points to problems and cascodes seem to be problematical.

Current tuning - this is with the tube CCS at the bottom replaced by a resistor and a 1.5V balanced input.

Screenshot 2021-09-14 at 06.19.46.png

And using one side of the amp output:

Screenshot 2021-09-14 at 06.21.08.png

At some point I will work out how to trick LTspice into FFTing a difference rather than only FFTing the node points.
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 14th September 2021 at 05:47 AM.
  Reply With Quote
Old 14th September 2021, 08:27 AM   #177
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
If I do this old school with the load lines it looks like the 12BH7As are struggling with the 320V and running lower in their voltage range.. if there's anything that tubes like is voltage.

So I've increased the rail voltages to 365V. I'll need to check the cathode heater voltages later but for now I want to see/describe the approach I've taken.

First. I want to run a tube CCS so that, if I'm running a single section, limits me to 20mA max and preferably 18mA total through both of the differential sections.

12BH7A-STR datasheet attached.

First I've put in a resistor rather than the CCS - we can add that back in later. -360V means it now has a lot of room down there.

From my understanding a cascode is a transconductance amp cascading into a transimpedance amp.
The input tube is a transconductance (converting voltage to current).
The 'cascode' tube with it's common grid then 'amplifies' the current by changing the resistance. My understanding this is a transimpedance - (converting current to voltage) and hence with Ra you then have an output voltage.
The next stage in the amp is a cathode follower - this, through a cap and 1M grid leak, presents a high impedance to the high impedance output of the cascode.

So in my head I take the input tube as a normal triode, although it's limited to a 9mA bias. The kicker here is - what is the Ra that the input tube sees as this defines the load line?

Well I look at this as 9mA through the input tube, then I have the same 9mA idle that needs to go through the cascode as a operating Q point. This means that both tubes should really run with a voltage across the tube to handle the current. The lower tube sets the voltage bias point so you're simply adding to that.
So you know the bias point, the required idle current and so you're then free to choose the upper voltage and Ra. Within reason.. the tube needs room to breath in voltage, this means the rail voltage increases (hence why SS folded cascodes are popular).

Input triode
Input range is 3.16*1.414 ie 4.5V but the reality here is to run with a cathode bias of 6-7V. The data sheet suggests -10.5V and 11mA. Instead to run a Iq of 9mA and a 6-7V cathode bias means our Q point is about 175V.
This means the differential pair is drawing an idle current of 18mA.

I'm positioning the triode to be able to take a direct input through a 4.79K resistor. There's no DC blocking cap in the way.
Also my GNFB would also feed into this (I could feed into the drivers but for now I'm running without NFB.

Cascode triode
Now we know the voltage across our input should be 175V, I've added 6.7V to that to give 182V. which is now the target for the top of the input triode. I can set that by using two diodes in series of 150V+30V. In the sim this results at about 186V. Near enough.
What I'm doing here (rightly or wrongly) is attempting to match the current at idle so the input and cascode are running at 9mA. The final output of the cascode will obviously depend on Ra above the cascode tube.

Next I add 186V+175V to give 361V. Now the Ra of 500ohm means 0.009*500 = 4.5V drop, so 361+4.5 = 365.5.. so hence the 360V rail.
The resulting output is 4Vpp with an 1.5Vp input.


Driver triode cathode follower
The driver triode is more like a buffer, presenting a high input impedance but low output impedance to provide current drive into the paralleled tube sections.
The standard M60 uses drivers to direct connect and bias to the output cyclotron pairs. My design is slightly different that uses a 1uF+0.22uF cap to provide enough drive to the four parallel grids.
This also has the benefit of being able to use the full 20mA range plus it's connected rail to inject noise from both rails. The downside of this technique is that it now has a large resistor in the 20mA max flow.. which is looking at 3W so parallel resistors to reduce to 1.5W would be needed to stop the thermal noise levels increasing.


The H3 sim output is annoying, however from others - this is often less evident in the real amp.
Attached Files
File Type: pdf RT043.pdf (320.5 KB, 3 views)
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 14th September 2021 at 08:46 AM.
  Reply With Quote
Old 14th September 2021, 04:54 PM   #178
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
Just starting a longer run following an adjustment of the cascode and then connected the balanced GNFB..

Screenshot 2021-09-14 at 17.51.18.png

Still got a while to sim.. but I think I may have overdone it with FB.. output is 1.5V peak into 55ohm.

Also need todo a bode plot again at some point.
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)
  Reply With Quote
Old 14th September 2021, 07:57 PM   #179
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
10s sim time, 1-10 FFTd:

Screenshot 2021-09-14 at 20.56.19.png

No I'll have to check the stability, ... and surprise surprise..

Screenshot 2021-09-14 at 21.07.20.png

So I'd better ease off the FB
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)

Last edited by NickKUK; 14th September 2021 at 08:08 PM.
  Reply With Quote
Old 15th September 2021, 06:01 AM   #180
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
 
Join Date: Dec 2019
So the tuned cascode front end is giving me 33dB of gain from 22 mu tubes without negative feedback.

Screenshot 2021-09-15 at 06.47.35.png

It's not as flat as my previous tuning, however this presents the harmonic spread shown before above.

The driver stage is attenuating - this has a potential divider (3Meg and 1Meg) on the grid to keep within the respectable grid parameters:

Screenshot 2021-09-15 at 06.50.34.png

The grids of the output ec99s see a little bottom end roll off:

Screenshot 2021-09-15 at 06.52.32.png

However this is first view of the problem that causes the large phase shifts when feedback is used:

Screenshot 2021-09-15 at 06.54.51.png

Green - the grid as before, the yellow shows the cathode of one of the triodes, and the red shows the respective load connection.

When NFB is used this amplifies this shift cause it to repeatedly shift the signal on each loop.

So as phase shift and group delay shift is related to resistance and capacitance configurations, I know the characteristics but not the location. I do know that changing the caps in the tube output changes this but the next diagnostic check is the solid state interaction with the caps.
__________________
V1 - low gain tube HPA OTL, fully balanced cascode 12BH7A, 12BH7A CCS, 12BH7A drivers and ECC99-BJT hybrid circlotron. (build)
V2 - pure digital tube HPA, RF-tube DSD512 modulation with mosfet output stage. (initial design)
  Reply With Quote

Reply


Designing my headphone ampHide this!Advertise here!
Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
I'm designing a TPA6120 based headphone amp, is there anyone can give me some advice shadownee Headphone Systems 13 31st August 2012 04:50 AM
C&C request: designing a headphone amp Thalyn Chip Amps 0 14th April 2012 12:12 PM
Designing a super low distortion headphone amp. Rainwulf Chip Amps 41 1st April 2011 12:24 AM
designing horn for compression driver & tweeters jayam000 Multi-Way 0 15th April 2009 09:26 AM
Designing for guys who like designing Sch3mat1c Tubes / Valves 14 31st August 2003 11:46 AM


New To Site? Need Help?

All times are GMT. The time now is 10:47 PM.


Search Engine Optimisation provided by DragonByte SEO (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Resources saved on this page: MySQL 13.64%
vBulletin Optimisation provided by vB Optimise (Pro) - vBulletin Mods & Addons Copyright © 2021 DragonByte Technologies Ltd.
Copyright ©1999-2021 diyAudio
Wiki