Constant Current Source (CCS) For Audio Applications

[Mark Johnson]

Would tying the gate to the supply pin on an Sk170 count as a single part?

Yes. If you've got a 2SK170-BL then its datasheet guarantees the current will be somewhere between 6mA and 12mA, a range of 2.00x.

The E202 current regulator diode datasheet guarantees its current will be somewhere between 1.68mA and 2.32mA, a range of 1.38x.

 

[thimios]

Using an improved CCS is wholly dependent on the topology it is used in.

Ranging from a "blameless' , where the CCS's are working with NFB in both
the LTP/VAS.
To something like the Groner TIS .... where the TIS does all the ripple cancellation.
CCS type (or lack of one) , has little effect here.
Or , an ultra high feedback circuit (Sansui) , where global NFB rules. No CCS
still gets you >-100db PSRR.

OS

Glad to see you here Pete!

 

[Mark Johnson]

Same parts count, higher impedance @ very low freqs & DC

replace 10K with 2N4117

_

 

[mchambin]

I arrived here, looking for measured data about CCS temperature stability.
Measured temperature coefficients of Bjts vbe and Leds.
I am interested in these two:
Led + BJT chosen for tempo cancellation.
BJT + BJT for a well known negative tempo.
Super high source impedance is not that important. For most applications fairly high and constant is enough.

 

[Mark Johnson]

Temperature compensated LM334 requires 4 components, figure attached.

The more traditional Opamp, Precision VREF, MOSFET, and current sensing resistor requires 5 components if you allow ~4V drop across the current source; 7 components if you only allow yourself to drop < 400mV across the current source.

~

 

[mchambin]

Searching over the net, I found:
Bjt Vbe at -2mV/C and -2.2mV/C.
Infra red led at -1.2mV/C and -1.6mV/C.

 

[Mark Johnson]

MAX6162 voltage reference IC is ±0.1% tolerance and tempco is 5 ppm/°C ; qty=1 price is $2.89

Vishay Dale PTF resistor family is ±0.1% tolerance and tempco is ±10ppm/°C ; qty=1 price is $0.92

Microchip MCP6V26 opamp input offset error voltage is 2 microvolts ; qty=1 price is $1.61

 

[mchambin]

I'll make my CCS for 1mA constant current with:

A cheap IR led ( a 50mA that do VF = 1.3v @ 20mA )
A 2N5551 with a 475 ohm emitter resistor. Very small parasitic cap, 160v range.
Biasing the led at 2.5mA ( with a 20k resistor to 50v ).
Bullet proof simple and cheap.
This will be good enough for my application, I bet I have better than 5% constant current over 0°C +60°C.

 

[Mark Johnson]

How many parts-per-million of error, per degree Centigrade, have you decided is "acceptable" ? In other words

What are the numerical values of Good Enough when you say

• "This will be good enough for my application"

 

[mchambin]

0.2%/°C
I do not need much accuracy, this CCS will be part of a biasing circuitry for a Class aB amp, where there are much other sources of inaccuracies.

 

[Mark Johnson]

0.2%/°C = 2000 ppm/°C

Indeed as you say, this is "not much accuracy". You will have an easy time building laboratory models and measuring them yourself, as you demand in post #44.

 

[mchambin]

I was asking for measured data because I failed to find consistent data about Led and Bjt in data sheets or articles about temperature coefficients.
No luck, but running in circles on copy pasted unverified internet rehashed data.
Meanwhile, I hope I can get reliable data with LTspice and Spice models.

 

[AKSA]

Mark,

Do you know how to design a CCS which reduces current with increasing temp?
Do you have suggestions?

I'm designing an amp which has a galloping quiescent with heat and I would best control it with a two terminal CCS.

HD

 

[mchambin]

Yes, that is the CCS made with 2 BJTs.
It has a worthwhile negative TempCo.
I do not remember the figure but I can dig out a Spice simulation.
Marc.

 

[mchambin]

I just simulated a CCS made of two 2N5551
With 10k to 50v biasing, 147 ohm emitter resistance to get a 5mA output.

A constant 5ma source over 0.9v to 100v
LTspice simulation results:
Ouput impedance is 5 Megohm.
Temperature stability is 0.5mA decrease over temperature increase from 10°C to 60°C
That is a TempCo = - 0.2% / °C

 

[AKSA]

Thank you, M. Chambin! I appreciate this very much.

I will examine my LTdoodlings and figure it out for my circuit.

Ciao,

Hugh

 

[mchambin]

5mA CCS

5mA CCS

 

[Elvee]

This is not really a two-terminal CCS, unless you are ready to accept unpleasant tradeoffs.
Here are some examples of 2 wire CCS, some of which can be designed with a -0.3%/°C tempco:
Improved 2W current sources (II)

 

[mchambin]

This is not really a two-terminal, unless you are ready to accept unpleasant tradeoffs.

What "unpleasant tradeoffs ?

Here are some examples of 2 wire CCS, some of which can be designed with a -0.3%/°C tempco:
Improved 2W current sources (II)

The circuit I proposed is simple, costs about $0.40 is -0.2% / °C tempco, is 5 MegOhm output, is 140v able from 0.9v to 160v.

An interesting detail about using it as a temperature sensor: The temperature negative tempco comes from the transistor that is not at the output. That is the one to thermally bound to the device one wants to monitor.
The transistor at the output is responsible for a negligeable positive tempco ( 70 times less than the transistor doing the positive tempco ), so leave this one at ambient temperature ( on a small heatsink, if necessary. Its self heating can be hight and depends of the voltage output of the CCS ).

The sensing transistor has a very low self heating because of a very low power, which furthermore is constant over all output voltages of the CCS.

I only see advantages in the CCS made with 2 BJTs ( at posts #55 #57 ).

 

[Elvee]

What "unpleasant tradeoffs ?

Your circuit is not natively floating or two-terminal, because of the auxiliary 50V supply tied to the ground.
To make it 2W, you need to remove V1 and attach R1 to the collector of Q2.
Try it, and you will become aware of the tradeoffs this implies...

The circuit I proposed is simple, costs about $0.40 is -0.2% / °C tempco, is 5 MegOhm output, is 140v able from 0.9v to 160v.

It is just the textbook 2-Q CCS, and its properties and limitations are very well known