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
Thread Tools
Old 8th March 2021, 09:48 AM   #91
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
The diode in the model is a MUR460, a 600V 4A If, 1.8A Irr device, however they don't give the Qrr value, the switching time is 75nS. (data sheet: https://www.mouser.co.uk/datasheet/2..._D-1773685.pdf)

Now if you look at this: https://www.mouser.co.uk/datasheet/2...00-1512439.pdf it's a 600V 3A forward but a 1.2A Irr, Qrr=17.5nC and the recovery time is around 9nS.

Better reverse characteristics but loose 1A forward. Given we can limit the current forward and the current is <3A anyway this isn't a big issue. Costs 0.20 more. In theory.. it can also be mounted on a heatsink with a max dissipation of 32W.
  Reply With Quote
Old 8th March 2021, 05:51 PM   #92
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
Screenshot 2021-03-08 at 17.38.07.png

That's a spicy meatball!

So this method is looking very very good indeed. Although the current may be a little weak on the eval board it goes to show that the SMPS could be tamed in terms of noise.

Tube amp he said.. may cost more than anticipated he said..
  Reply With Quote
Old 8th March 2021, 09:59 PM   #93
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
Slightly longer along the same model.. the amp is stabilising (about 4.5s in):

Screenshot 2021-03-08 at 21.50.33.png

So this is looking very good (there's a 10Khz on one channel and a 30Hz on the other to check for cross talk). The input signal is 447mV so the amp needs a little adjustment but I'm more interested in the power supply at the moment.
  Reply With Quote
Old 9th March 2021, 05:36 PM   #94
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
So after 500hours of cpu time, the model is getting a little unwieldy from LTSpice's bad memory management. Although the raw file is only 42GB, that's 6.2 seconds of simulated amp time with the switched mode power supply.

So time to start exploring hardware for the SMPS. The eval kit is over priced and underspecced and the other SMPS don't offer the same large cap support needed really.
  Reply With Quote
Old 13th March 2021, 10:00 AM   #95
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
Some one shoot me for going OTL lol.

Normal valves have voltage but low current, normal solid state has low voltage high current. OTL has both - high voltage and high current.

Power wise the amp is relatively simple - about 0.5A @ 200V (100W) B+ and then add about 2.5A@12.6V (32W).

To get clean power.. is when it becomes more difficult. Passive filters and linear power supplies throw away a lot of power. Rectification of AC and selecting the right rectification to not to loose your noodle, err I mean current (0.6x for full bridge) or the variants thereafter.. then your passive low pass filtering to make that noisy beast starts adding cost - the more you add the more more initial power is required.. in a ever decreasing spiral of return whilst ever increasing spiral of cost.
You want 1Vpp ripple ... that's a good 6mF and 5.1ohm resistor.. but that drops your voltage a little, and given you have 100-200W that a spicy resistor that's high cost and then add the cost (one 6mF may work but paralleled 6x1mF may be better).. and replicate that through the design to get sub 1Vpp.. that means more initial power.. and the spiral continues.
Oh.. that doesn't include inrush.. so the wife complains when the amp is switched on - the lights flicker and when the electricity bill arrives.

Nobody told me it was easy.. so suck it up.

SMPS.. 648W PSU 150mVpp ripple at 24V means high current, 27A to be precise.. that then needs handling from 25A@24V -> 2.5A@200V or a nice juicy 500W. The SMP process means you're speccing components that can handle 900-1.2kV to cope with the spikes at that current. In reality they're not going to be putting out 1800W+.. far from it.. but the current at 24V means the PCB traces need special attention, bigger to reduce resistances and help dissipate the heat they generate.. Heatsinks please!
So on the HV side, that diode.. 1.2Kv and fast switching - on a heatsink, that 5.1ohm resistor to make the main res capacitance a 1st order 0.5Hz LPF - making the SMPS noise *elma thud*verwy verwy qwiet now has to cope with not the draw of your amp.. but the full startup current you've programmed into the 24V-200V cap-charge SMPS. Now that 5.5milliohm sense resistor is now giving a good 1.9A@200V on the transformer side - 400W for a couple of seconds - although after it can kick back and relax as the system sees 1/4 of that. Heatsink please!
Then you have the rez caps .. well you could split this, the caps closer to the stages lower current for the resistors (1-10W).. 450V caps here too.. the down side in reducing the main res LPF capacitance is that you get more noise closer to the tubes. 5.1ohm + just a single 1mF is 28Hz with about -10dB at 100Hz.. however on the plus side this is expandable as we go on with 1mF caps being able to be added as an upgrade and the other 750uF/1mF caps closer to the stages help transient response and lower noise generated within the amp.
At least I'm starting with 150mV ripple.. active power correction, 70%+ efficiency with some noise 70KHz+ that is easier to filter out. All thanks to that little cap-charger/regulator.
I'm considering a little bit of a 'adjustment' simply because it may be a little overkill to supply 600+W simply to charge 6mF of capacitance in a couple of seconds after which it's only running 1/3 of that. The heaters need 11 seconds for B+, so in theory I could make the amp current simply use a little less current.
If I increase the cap-charger sense resistor I can drop the current limiting, 5.5mOhm for 25A / 2.5A max, which is about 1.9A or so.. could be dropped to 1A on the 200V side. It may take couple of seconds longer to get to B+ but at the same time that soft ramp means the tube heaters and the B+ could be made to go hand in hand..
Who'da thought you could make a SMPS behave like an old school tube rectifier (ok that depends on the type.. some tube rectifiers conduct immediately).
It could also mean a smaller 10A of 24V.. which means a smaller power supply.. (I'd still keep the same charge controller switch components) and a smaller LPF power resistor too!

All about 88% efficient.. power factor of 0.95, regulated, ....
Meanwell MSP-600-24, 25A 139
Meanwell MSP-450-24, 18.4A 116
Meanwell MSP-300-24, 14A 86

A 24V to 12V supply for the heaters - well it's actually cheaper to get a 100W MSP-100-12V rather than try and convert to 24->12V. The benefit of this route is that you get a 8.5A of 12V, regulated, power efficient.. active PFC... and (at the moment the data sheet isn't available) being the MSP-100-12.. you get that same ripple Vpp to help reduce the noise. 37. The PSUs are adjustable - up to 13.something V. 41 and you have a 15V version.

So the idea is that the system is likely to have a 300W and a 100W power supply. The medical models from mean well are also fully isolated, you can even parallel them and they also have loss sensing so you could make it a feedback around smaller regulator.. but lastly - you can tie the V- to ground for reference (or any voltage should you wish!) that means the heaters could be elevated too. not that I'd expect hum.
A simple current regulator on the heater supplies should prevent hot spotting at start.

The supplies shouldn't take too much space and will connect to the same AC inlet.

120+VAT (20%) is a lot. However it gives me 150mV ripple as a starting point, and when compared to the linear PSUs it's probably cheaper. I still have the for the cap charger board and the rest of the amp but to be honest, I'd have needed a lot more extra for fixing the inrush for the linear.

Anyways.. onwards and upwards
  Reply With Quote
Old 13th March 2021, 06:34 PM   #96
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
I've been investigating this more.. before I press buy.

The LT3751 has two modes - charge and regulation, with no-load being ignored.

So looking at the model I'm using a 5.5mOhm resistor to set the current limit, which in turn pretty much dominates the power output. So I'be been seeing if this would fit into the 300W as mentioned above.

The chip has a number of modes and sub-modes. The main being charge and regulate. However the approach also varies between the modes;
* charge, peak current for charging until either the current drops or the FB gets to the desired level
* regulate, has it's own peak current and the loading approach is split depending on the FB level:

Charge 100% Imax duty cycle 100% current used
Regulate 10-95%
* Heavy 50-95% Imax duty cycle
* Moderate 10-50% Imax duty cycle
* Light = 10% Imax
No-load = Imax where voltages rise above the FB>1.34V IIRC.

Regulating heavy load, the peak current current can vary up to 95%, but the period between pulses varies. the lighter modes are simply a way of saying it varies both the amplitude of the peaks (<95%) and the the number of peaks per 26KHz clock (ie in 38uS)..

In heavy load, the power output is:
Pout = 1/2 * Ipk / (1/Vtrans + N/Vout)

Interesting if we set our Ipk to the max 25A for 24V, with our 1:10 trans, for 200V we get:
Pout = 1/2 * 25 / (1/24 + 10/200) = 136 Watts - this is our maximum that the transformer can cope with.
As P=VI, 136/200 = 680mA at 200V being the absolute maximum ignoring reality.

In charge mode, if you want to ensure a charge in a given time period then
Ipk = (2*N*Vtrans+Vout)*Cout*Vout / efficiency*Vtrans*(Tcharge-td)

But this differs for regulation, where Ipk is
Ipk = 2*Poutavg/efficiency * (1/Vtrans+N/Vout)

Now assuming efficiency at 70%, that's 0.7. So the formula would read:
25 = 2*Pout / 0.70 * (1/24+10/200)

25*.5*.7 / (1/24+1/20) = Pout = 95.45 W.. so at 200V that would be 477mA.

So let me get this right.. 25A * 24V = 600W. And out of that 600W you get 95W regulated.. or 15% efficient.

Well.. that would depend on the period of the pulse and how many times/sec that is.

They give these two formulae..

Lpri = 3uS*Vout / Ipk*N
Lpri < 38uS / Ipk*(1/Vtrans +N/Vout)

Note that if both IPK and N are increased signifi- cantly for a given VTRANS and VOUT, the maximum IPK will not be reached within the refresh clock period. This will result in a lower than expected maximum output power.
They give a test for this scenario..

Lpri = 3x10^-6 * 200 / 25*10 = 2.4uH
38x10^6 / 25*(1/24+10/200) = 0.1393uH
So there's a possible problem because this is not true 2.4uH < 0.1393uH !

So basically this means we can't deliver enough power per charge spike. We would need to increase the Vtrans, instead of 24V this would need to be higher. However the transformer has a max of 24V. The chip can do Vtrans of 40V but then that messes with the power supply which would need to be 48V with something limiting the voltage to 40V.

This was going soooo well but I suppose you can't count your chickens until they're hatched..

This would explain the reason why they only get 200-300mA out of it in regulation mode. So in order to be useful I'd need two - one for each channel.

Hmm time to lick my wounds and have a rethink.
  Reply With Quote
Old 15th March 2021, 10:51 PM   #97
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019

a) I will be using a isolated, PFC, 150mVpp 24V medical PSU.. it means I'm already isolated and filtered from the mains.
b) Flyback has issues.
c) Boost is a simple option...

I've been looking at the LT1243 which is a 50-100% duty cycle PWM chip that has feedback voltage and current sense. It also has a 'programmable' oscillator with PWM pulses then following the external RC network. This means you can control the RC and even modify it's behaviour based on load etc. 1MHz maximum PWM is not to be sneezed at.. so given the flyback didn't work well with the chip I went down the boost single mosfet approach.

So it gives me close to 130W (about 180V at 600mA+)in it's current configuration with a 280R resistor.. so time to try it on the amp model (ie running 447mV input).

So far the B+ lines are at 200V, and the system has started to backoff 'blanking' cycles from the PWM. At 5secs the amp switches off the negative startup bias for the tubes. So we should see that B+ voltage dip back to about 150-180V.

Screenshot 2021-03-15 at 21.41.33.png

Top is the B+ line at the output tubes.. bottom is the power being pulled from the 24V power supply. You see the behaviour of blanking PWM to right as the power draw is backed off. That 400W is the peaking off the PWM, so in reality is less than that.
  Reply With Quote
Old 17th March 2021, 07:40 AM   #98
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
Current run is looking very good, noise levels are low, it's hitting 150V at the output tube and is taking peaks of 130W (probably less averaged) without dropping to idle. So this is looking good for a 300W supply.

This model is the full amp (heaters are running off a separate supply) but with 447mV input signal. I want to run a 3.16V +20dBV sustained output just to see the behaviour (if we see any sagging etc). I don't foresee a problem with current and power but just checking is worth the time. At that level all the tubes are running at max signal (ie output signal extends to close to 60mA per triode) for full power.

It's worth spending a little time characterising the design to ensure it's going to behave in the right way at different power levels (0V input, normal 447mV and max 3.16V).
Given the there is a bank of caps on the HV side, it's worth looking at inrush at the start as the 24V from the supply will find a very low impedance path through the HV diodes and then through the 5.1ohm resistor. Not that the design can't handle that - I have very little doubt about that but the 24V PSU may trigger it's protection given the immediate current draw.

I also have to think about the same for limiting the current to the heaters. Otherwise the heater SMPS will simply see a short circuit until the heater warms up.

An FFT from 7-8.1s of the simulation after startup..
Screenshot 2021-03-17 at 06.41.22.png

So that's showing the 10KHz input on the left and a some cross talk from the 30Hz input on the right channel.

This is the model it's running:

Screenshot 2021-03-16 at 08.57.33.png

Last edited by NickKUK; 17th March 2021 at 07:44 AM.
  Reply With Quote
Old 17th March 2021, 08:52 PM   #99
PRR is offline PRR  United States
diyAudio Member
PRR's Avatar
Join Date: Jun 2003
Location: Maine USA
Designing my headphone amp
> I want to run a 3.16V +20dBV sustained output

+10dBV. (20dB above -10dBv, a common hi-fi reference from days of tape.)
  Reply With Quote
Old 18th March 2021, 02:59 PM   #100
NickKUK is offline NickKUK  United Kingdom
diyAudio Member
Join Date: Dec 2019
Originally Posted by PRR View Post
> I want to run a 3.16V +20dBV sustained output

+10dBV. (20dB above -10dBv, a common hi-fi reference from days of tape.)
*thumbs up* You're right - I should recap my notes:


+20dB above -10dbV it is!
  Reply With Quote


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 05:50 AM
C&C request: designing a headphone amp Thalyn Chip Amps 0 14th April 2012 01:12 PM
Designing a super low distortion headphone amp. Rainwulf Chip Amps 41 1st April 2011 01:24 AM
designing horn for compression driver &amp; tweeters jayam000 Multi-Way 0 15th April 2009 10:26 AM
Designing for guys who like designing Sch3mat1c Tubes / Valves 14 31st August 2003 12:46 PM

New To Site? Need Help?

All times are GMT. The time now is 12:59 AM.

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