A Pass inspired Sk170 Headphone Amp - is this stupid

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Razor,
That won't work very well. To prevent oscillations you need a gate stopper on every JFET's gate, and to prevent current hogging you need a resistor in every JFET's Source. You need a resistor to discharge the output and input cap. Also, you need a fillter on voltage divider at JFET's gates - battery is not so clean as one might think. Also, the CCS (Q5) to supply the LED Vref should be used. Five instead of four Source Followers JFETs is better (10 is even better).

You don't write what is the grade of your JFETs (GR, BL or V) but use the highest Idss parts you got and set the P1 so that you get about 80-90% of Idss through every Source Follower JFET (measure the voltage drop on 4R7 Source resistors).
If you got, for example, JFETs with Idss=10mA you'll wan't to run about 9mA through each of them, so the Q6 CCS should supply about 40-45mA (if you use 5 JFETs as in my sch.). The LED used should be nothing fancy, ultra bright etc., just plain old, 3mm red one. If Q6 gets hot, use the bigger one - BD139 or similar.
Viel Spass !

Edit: there's an error - the resistor in Q7's Drain shouldn't be there, that Drain should connect straight to +12V


Juma,

thank you for your input. I see things a little more clearyl now Idss of my fets is mostly 7ma-8ma, so i will aim at 6.3ma Bias . I see you use one SK170 as current source for the LED in the voltage reference and i guess the little trim pot is to adjust bias current. I will implement this. So no soldering tonight, back to the drawing board it is.

Thank you all for your help und besonders vielen Dank an Dich Juma
 
So, a long day has passed. I took all your suggestions and here are the new schematics.

Still we have a basic Common Drain Jfet Amplifier, biased by a constant current source, i will go through each part an what it does, this helps me to learn, please correct me if you see anything wrong, this helps with the the learning process.

V1 is the Voltage source, a 12v lead acid battery, i would not hurt to put a storage cap (1000uF) here
R7 R8 and R2 form a voltage divider to bias the gate to 1/2 supply voltage
R7 and C5 form a RC Filter to get clean voltage to the gate
R1 is the input Resistor and also draws current from Cin
Cin removes any DC input to the Amp also Cin and R2 form a CR filter so i have to be careful since this will have an effect on the frequency respose of the Amplifier
J1 to J15 are N-Channel Jfets these will provide the current gain needed by the headphones, they are biased at 90% Idss, in my case that IDss is 7ma to 8ma so i will go for 6.3ma bias. At high input voltages this will results in slightly positive Vgs but thats not a problem as long as i keep it below 100mv (according to Boberlys papers)
Rg1 to RG15 are gate resisitors they insure stability and prevent HF oscillation
Rs1 to RS are Source resistors, they prevent current hogging and ensure that the same current runs through all Jfets, the also provide a handy spot to mesure bis current
J16 R3 R4 C2 Led1 R5 and Pot10 form a current source, we need it because we bias the Jfets via a constant current
J16 is used as a CCS to provide current to the Led the voltage drop across the led is used to bias Q1
R4 and C3 also form a CR Filter to clean ub the bias Voltage for Q1

QUESTION : why is R4 220 Ohms? Simulation shows this will only give about 1.2ma current to the LED shouldnt there be at least 10ma or more through the LED? Is the value correct or would 22 Ohm be better? And does R3 only functions as a gate resistor?

R5 and Pot 10 set the current for the current sink, the pot helps to set the current exactly while measuring the voltage over RS
Cout1 Blocks the DC Voltage from the Headphones
Cout2 Is there because Elkos are bad and you need to bypass them, on the other hand that can cause oscillation problems, further investigations and listening tests have to be done here.
Rbleed draws the current from Cout1
and finally
Rload simulates the headphones

QUESTION: Why am i doing this, this is totally stupid, why not just a power mosfet for Euros, save myself a lot of trouble
ANSWER: Because i learned so much in the past two days. This is a fun project, i will never learn anything if i just build kits, i want to understand how these things work, not just wiring two opamps together like in a schematic found somewhere on the internet and this unused 2Sk170s finally get some use.

I am grateful for the support of the forum Elders, this is a great community and hopefully this will be a great Headphone amp, if not at least i learned a lot and that is whats most important for me.
 

Attachments

  • jfetheadphone3.jpeg
    jfetheadphone3.jpeg
    88.5 KB · Views: 518
... QUESTION : why is R4 220 Ohms? Simulation shows this will only give about 1.2ma current to the LED shouldnt there be at least 10ma or more through the LED? Is the value correct or would 22 Ohm be better? And does R3 only functions as a gate resistor?...

LED is just a voltage reference element for the CCS so we don't need a lot of current through it to does its job - 1mA is enough for the sch. I drew.
But if you plan to run about 100mA through Q1 and if we assume its hfe is about 100 it would be better to run 4-5mA through the J16 since its current will be shared between the LED and Q1's base. In that case C3 can be of higher value (470u - 1mF) and Q1 should be BD139 or similar.
Yes, R3 is just a gate stopper.
 
Its not a Headamp for MC Cartridges its a headphone amplifier. I dont want to make a deafining torture device. In the case of the Sennheiser HD598 we are talking about 0.1% Thd at 110dB spl, well in simulation at least.

Q1 will be changed to BD139 PD willw bw around 0.5W better safe then sorry.
R4 will be lowered to 22 Ohm resulting in 4-5 ma in J16
C3 increased to 1000uF to keep the corner frequency of the rc filter
A larger storage cap will be added to V1.

So, if everyone is ok with the circuit i will place the order for the missing resistors and stuff and start making a pcb, this is a little overkill for strip board.
 
Last edited:
Member
Joined 2010
Paid Member
some reading

Hi there,

Have a look at This which mirrors your efforts in some ways and discusses paralleling Jfets for effect.

There's also a great image that seems relevant detailing ccs layouts... found it! may be relevant again, and provided some good learning for me.
 

Attachments

  • CCS-Test-2.gif
    CCS-Test-2.gif
    38.4 KB · Views: 355
Well, along those lines, here's another sim'ed eval. I don't know where I got this or the conditions such as the applied sineV etc. But, the relative rejection should correlate with the previously posted data.
 

Attachments

  • ccs_collection.png
    ccs_collection.png
    7.1 KB · Views: 310
Its not a Headamp for MC Cartridges its a headphone amplifier. I dont want to make a deafining torture device. In the case of the Sennheiser HD598 we are talking about 0.1% Thd at 110dB spl, well in simulation at least.

Q1 will be changed to BD139 PD willw bw around 0.5W better safe then sorry.
R4 will be lowered to 22 Ohm resulting in 4-5 ma in J16
C3 increased to 1000uF to keep the corner frequency of the rc filter
A larger storage cap will be added to V1.

So, if everyone is ok with the circuit i will place the order for the missing resistors and stuff and start making a pcb, this is a little overkill for strip board.

And you know 0.1% THD will be good sounding??
BTW the PSRR will be no better with CCS for a led reference supply, but you already know that, you just like to stick as much jfets in the circuit as you can ;)
 
Howdy,

yes i build the one FET prototype, it works fine but really needs a very silent power supply. Its pretty straight forward and easy to build, but it never left prototype stage. If you dont have a scope its a great way to evaluate Power supply noise, just connect amp and headphones and listen for noise :)
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.