JLH Headphone Amp

The ground point is actually a 5mm round pad . Capacitors connect up on one side and the amp grounds connect about 120 degrees away from it.
However I just realised that one channel has 40mm of track common to the transformer 0V line .:eek:
I can fix that easily. Will do it and compare the FFT to see what it has done with and without a test signal.
Hmm....:scratch: Why was I so careless !
Thanks for bringing this up. I haven't yet checked square waves on the system or taken a frequency response measurement with real headphones connected.
The heatsink temperature is 57 deg C and ambient is 27 deg C with a quiescent current of 135 mA per channel. So at this current one will need a larger heatsink. In an enclosed box the temperature would be higher I guess. Might be a good idea to clamp the transistors to the casing.
 
Last edited:
Yup. That shared section made all the difference. It's dropped to about -76dB wrt to the output at 1 volt. That's close to a 20dB drop !
It's close to the 2nd harmonic level. I must try to get it down a bit more.

After that a few more tests before I get to listen to it again and see if I can notice the improvements. Of course the hum might not be audible now.

At 135mA per channel it isn't exactly battery operable. But at 35mA per channel and high impedance phones it could be used with batteries. Might chew up ordinary cells easily!

After chopping up the traces on this board and setting everything right I might make a couple of more fresh boards.I want to try out the MJE340 as they have much higher (measured) gain than the BD139 I have.

Cheers.
 
Administrator
Joined 2007
Paid Member
When you listen to it and are familiar with the sound try cutting the quiescent current down... you might be surprised how low you can go before it's a problem audibly.
Maybe have it switchable for battery operation... low Iq.
I'm looking at battery power... mine will only really be used as a test amp for evaluating stuff rather than long term listening. Some camera battery packs are pretty cheap nowadays, 12 volt 2200mah etc as they are becoming a bit obsolete with new low power solid state camcorders.
 
Yes you are probably right about reducing Iq to an acceptable level audibly.
I spent a lot of time just now trying to put up the performance curves. But for some reason I couldn't transfer it from the other computer ( which doesn't accept USB memory as of now ). I'll do it later.

I used a Sennheiser HD580 , Philips SHP805 and Philips SHP8900 . I think the 8900 sounds the best of them all and a fraction of the price of the Sennheiser.
I see that many guys recommend the new Philips 5400.

No more audible hum but I can sense the headphone is connected when it is plugged in. That must be due to residual hum though I don't hear it well enough to call it hum.

The LM317 has better performance ( -80 dB hum rejection ) compared to the LM7812 ( -60 dB ). So I'll try this later.
Right now it sounds good. There is enouigh bass to over load the 805's and the older 800's. I don't use any eq. It certainly is louder than I think is safe for continuous listening.

I'm certainly going to make more boards. It's a good gift for those who like listening to phones. Now it will be interesting to see what it does when used as a preamp ! Shall do that sometime soon. A smd input section might be a good idea ! No, I better avoid that. Too many things to be built/completed now.
Cheers.
( I'll post plots tomorrow).
 
Here are some measurements

Frequency response with 33 ohm load, distortion against output voltage , Ripple after ground modification,Linear distortion FFT and ripple before ground modification.
 

Attachments

  • 33OHMFRS.BMP
    37.6 KB · Views: 804
  • DISTOR-3.BMP
    37.6 KB · Views: 517
  • RIPPLE-7.BMP
    37.6 KB · Views: 377
  • 33OHM-LN.BMP
    37.6 KB · Views: 385
  • 33-OHM-1V.BMP
    37.6 KB · Views: 403
Last edited:
My current layout is not good enough. I have had to cut traces and add external wires to sort out the earth. Let me finish the modified layout. I'm trying to keep the pcb size down which is already about 80 x 100 mm .
The additional earthing traces require space or I will have to move the parts around and see if that can reduce the tracks. It's and interesting problem but will take some time.

I think it sounds very good. Certainly worth the time having built it. I think it sounds cleaner than the headphone amp I mentioned above. Bass is very nice on the 8900 phones. The other Philips phones overload on the bass notes ! Only the Sennheiser and the Philips 8900 handle the low end without any stress. You could always roll off the bass early by using a smaller input cap for phones that have low end limitations.Tremendous transients and good mids. Maybe I should say 'transparent sound' ? All this with that cheap yellow box 1uF input cap and Samwha electrolytics ! I'm surprised. I will try some Solen's at the input and see what happens.

I'll put up my current board design. Can't say how long it will take to get the modified one ready. I've a lot of other work to get done in the next few weeks.
 
I managed to make a Jpeg file of the board layout. It went through several conversions and the end result looks bad. However it isn't the final one and is just a record of what was done right now.
Cheers.
 

Attachments

  • JLH-pcb.jpg
    JLH-pcb.jpg
    52.3 KB · Views: 3,777
Some more results.

First of all I must say that I was wrong about the bass overload on a couple of phones that I mentioned earlier. The problem was because of mixed up input cables causing very low loading on the source.:boggled:

Now I have some new plots. The output impedance of the JLh phones amp. All these are measured results. The output impedance is lower than about 0.5 ohms and keeps rising with reducing frequency because of the coupling capacitor. This however doesn't cause too much of a problem as you can see from the frequency response plots with the headphone connected. I've shown the impedance plots of the headphones also. I've picked three of them. Two Philips units and one Sennheiser. That's from low impedance to high impedance. The Sennheiser has the most horrible connecting plug at the phones and I keep wanting to throw them out the window. NONE of the other headphones have any problems with connection including the Philips which has a plug at the phones. Can't believe that such an expensive headphone can have such a poor connection ! Guess they can get away because they are so big ! Did they ever do a recall ?...... like the Toyota Prius !

Since output impedance can cause ripples in the frequency response when it is loaded with varying loads like speakers or headphones, I tried a test with a 5 ohm resistor in series with the low impedance headphones ( about 40ohms). On a 1db/div plot I could hardly see any difference from the plot connected direct to the amp. The last three plots are with the headphones connected to the amp.
The plots are
1. Amp output impedance
2. Open circuit response of the amplifier
3. Philips SBC-HP800 phones impedance plot
4. Philips SHP8900 phones impedance plot
5. Sennheiser HD580 impedance plot
6. Philips SBC HP800 electrical response at the amp output.
7. Philips SHP 8900 electrical response at the amp output.
8. Sennheiser HD580 electrical response at the amp output.
 

Attachments

  • SBCHP800.BMP
    37.6 KB · Views: 306
  • OPENCKFR.BMP
    37.6 KB · Views: 322
  • JLH-ZOUT.BMP
    37.6 KB · Views: 538
  • SHP-8900.BMP
    37.6 KB · Views: 286
  • HD580-PL.BMP
    37.6 KB · Views: 334
  • SBC800R2.BMP
    37.6 KB · Views: 253
  • SHP8900R.BMP
    37.6 KB · Views: 274
  • HD580-R2.BMP
    37.6 KB · Views: 366
Capacitors and preamp use !

I used the JLH headphone amp as a preamp.
The power amp that I used sounds very good by itself. Great low end, tight bass and very clean and clear HF.

When I added the JLH amp the low end sounded different but good. HF seemed to get slightly dulled. Loss of bite in the HF !
The power amp already has a film cap at it's input. The JLH has an electrolytic cap ( 47uF/16 V Samwha ) at it's output. So out came the Samwha. Now the HF got much better and bass is still very good.

The difference between the bass ( drum notes) with the power amp by itself and via the preamp is that the drum seems to sound more powerful with the preamp.It is also better defined than without the preamp. This is not an amplitude issue. I know, I've measured the response and seen it has not changed. You can also see that at 20 Hz it's just 0.1dB down with an 8K ohm load. The difference is very audible even without an AB test.

I then changed the input cap from the 1uF yellow box cap to a 0.68uF ( all I had ) Solen cap. I expected a huge difference. I find about 10% difference ! By that I mean that it did improve the HF a bit but all else remains the same.
Some of the HF bite is back.
With the altered JLH amp the power amp sounds better though there is still a very slight loss of 'bite' in the HF. However voices seem to sound more real....tube like ? Maybe.
Try the album called "The Great Basso - Vol1 ". It's Chinese but I like the voice and great recording quality. Used as demo music by high-end shops in the Far East. As the title suggests the singer has a very bass heavy voice.

I wouldn't hesitate using this as a preamp whenever I need one. Sounds very good. However if you need to decouple the dc at the output ,you might have to use a good quality film cap. No Solen cap for the input ?.... the 1uF box is pretty decent. I must try out my low cost 1uF /250V polyester cap ! It's too late now. I have to crash out . It's almost 12.40 am .
 
Time for some drastic mods.

Today is a great day. Clear sky and 26 degC ambient !
Perfect for starting a new project ! This time it will be a split supply JLH headphone amp. I expect it to sound similar to the single supply version ...but can't be sure till I try it out.
Right now the 6Volts dc of the headphone amp is biasing the input capacitor of the power amp at a much higher voltage than the usual low mV across it if it had no dc at the input ( of the power amp ). Does it help ? We'll see.
Even the split supply circuit requires an input cap which could be a good thing in general as we can never be sure if the source is dc free. With a "good" coupling cap we could remove the output coupling cap in the DAC.
That eliminates two critical capacitors in the signal path.
 
Did some more listening tests with different input caps. From what I have I'd pick the Solen followed by the Epcos X2 caps and then the low cost film caps. I'll post a picture rather than descibe them. Top would be best and bottom would be least preffered. However I'd like to point out that even the bottom one sounds quite good by itself. If listened to with large time gaps I think it would be difficult to differentiate between them.
 
Here are the caps. Best one on top and going down with decreasing performance. Most are quite decent though the grey X2 cap at the bottom doesn't sound as good as the rest. The blue cap is an Epcos 1uF X2 cap.
Caps in the same horizontal line are fairly similar to each other.
I'd say the Solen ( 0.68uF) has a slightly fuller sound on voice and bass ( just a wee bit ) and piano notes are just a bit more incisive.
But the difference is not 'night and day' ! The nay sayers might say that there wouldn't be a difference. Well it doesn't matter. I'll just put the cap that suits the cost of the rest of the board and which doesn't 'sound' bad. That might be the yellow box cap !
 

Attachments

  • Input caps.jpg
    Input caps.jpg
    44.6 KB · Views: 2,747
Last edited:
I just got back from an out of town trip. I see that no one appears to have tried the JLH headphone amp after I posted some results.
Well I'm on to the split supply one and tried to sim it. It seems to be good too. Setting the dc offset was a small problem though I think I have a solution. Must see if it's stable with time.
 
Administrator
Joined 2007
Paid Member
I just got back from an out of town trip. I see that no one appears to have tried the JLH headphone amp after I posted some results.
Well I'm on to the split supply one and tried to sim it. It seems to be good too. Setting the dc offset was a small problem though I think I have a solution. Must see if it's stable with time.

It's nice having a separate forum for headphones but in a way I think it would be better in the solid state section.
 
Hi!
I built the first JLH amplifier with +/-12V supply.
In my opinion something's wrong with it. The heat sink becomes very hot within half a minute. The heatsink is quite big, much bigger than Ashok's one on the photo as I see. It eats about 400mA, that's too much, isn't that?
I am afraid it may die if runs too much, but I measured some part of the amplifier . I attached a picture. On it there are some value that I measured using a multimeter.
What do you think? Is there any problem with it?
Thanks.
 

Attachments

  • JLH.jpg
    JLH.jpg
    78.1 KB · Views: 2,979
Kenpeter is right. The voltage at the base of Q5 should be about 10.8 volts. You can change R11 to reduce the bias current in the output stage. Right now it appears to be about 0.110 amps . Additionally you must make sure the heat sink is well installed on the transistors. Silicone grease is a must on both sides of the insulating mica washer !
You must remember that in my circuit the supply is a "single" +12V supply. Voltage across the output transistors is about 6 volts each and so dissipation is "half" of what you have. You can drop the output current to just about 80 mA ( R 11 = about 6.8 ohms) to drop dissipation and you will still get enough output on a 32 ohm headphone to permanently damage your hearing ( over 110 dB spl ).

The base of Q4 also seems to be a bit high. It should be about -1.2 volts. R1 and R2 voltage drop will reduce this voltage a bit ( sim value -1.18 V ). Diodes oriented correctly ?

Do you really get 0V at the output with the components shown ? The sim shows +0.904volts ! R7 and R4 determines the output dc voltage. R7 is supposed to drop the difference of the two diode drops - (R2+R1) drops - Vbe drop of Q4. Any missmatch here will cause an error. R4 could be adjusted to get 0 V at the output. Note that R7 and R4 have the same current flowing through them (approximately). But R4 has 1.2 volts across it. So R7 will have too much across it to get 0 volts at the output. Alternatively D1/D2 should actually have one more diode in series with them! Easiest method I think is to adjust R4 and/or R7 to get less than 1 mV at the output. We will have to check how much the dc drift will be with time.