# current mirror supplied with charge pump

Discussion in 'General Electronics Chat' started by kevin0228ca, Jun 18, 2015.

1. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0
I designed a circuit with current mirror supplied with charge pump.
Current mirror circuit and charge pump circuit each work fine by themselves,
but when I supply charge pump voltage to current mirror voltage source, my signal got weird.

Current mirror
My Rload is varying and I need constant current.
mirroring a 5V square wave by R1=500 ohm, giving 10 mA.
I tested with 12V supply for Rload < 1k and got about constant 10 mA square wave.

Charge pump
since my Rload will be > 1k, could be 5k, I need more votlage supply so I use a charge pump.
I tested my charge pump with 12V supply and got about 40V

So I supply charge pump to current mirror, but got weird signal at Rload, which was working when supplied with 12V.
Also, Vsupply went from 40V to about 30V.
Simulation works though, but I do get spikes of current between square wave 5 and 0 V.

Can anyone please check my circuit?
Am I using correct power parts?
And since I am new electric engineering, is my circuit good design?
Thank you.

fE is just uF

• ###### circuit.asc
File size:
5.9 KB
Views:
4
Last edited: Jun 18, 2015
2. ### crutschow Expert

Mar 14, 2008
13,000
3,229
Your circuit is a strange mishmash of designs.
Why do you need 40v? The load takes 15V.
What is the purpose of the current mirror since you already have U1-M1 as a constant current source?
The mirror doubles the current drawn from your charge pump.
R2 is a dummy load and not needed. It just wastes 40mA of current from the charge pump.
How can fE be μF?
Post the .asc file.

3. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0

I forgot to mention Rload will be varying > 1.5K, could be 5k, so I need more voltage and current mirror to keep constant current.
So I use u1 to mirror constant current.
Is my method correct?

How to know maximum current of charge pump?
And should I test current of Vsupply to see if exceed?

R2 I forgot to erase from circuit from where I was only testing charge pump.
fE is uF because I don't know why my ltspice is showing fE when I indeed enter uF as capacitor, I think just a bug of character.

I want to upload .asc file but saying uploaded file is empty

4. ### crutschow Expert

Mar 14, 2008
13,000
3,229
So you want 10mA into a 5kΩ load? That requires 50V across the load and your charge pump is only supplying 40V.
A current mirror doubles the required current. Modifying your constant current source to drive a grounded load might be better.
You click "Upload a file" and then select the xxx.asc file from its location on your computer. Is that what you did?

Last edited: Jun 18, 2015
5. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0

I simulated I could get 50V if I increase my diode and capacitor in my charge pump.
I am only testing with loads < 4K now, so my breadboard I only build charge pump to 40V.
could there be problem if I have too much multiplier?

How should I modify my current source for a grounded load?
Is there a example?
Do I still use a current mirror?

File size:
5.9 KB
Views:
2
6. ### crutschow Expert

Mar 14, 2008
13,000
3,229
Okay, I've got your circuit working.
But it's late here so I'll get back with you tomorrow.

7. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0
Thank you, meanwhile I test currents and voltages

8. ### AnalogKid Distinguished Member

Aug 1, 2013
4,529
1,248
If the input always will be pulses rather than some other waveform, then you can create a fixed 10 mA current source and just turn it on and off. Note that the semiconductors might have to be adjusted for voltage margin when operating at 50 V.

ak

9. ### crutschow Expert

Mar 14, 2008
13,000
3,229
How accurate and stable do you need the current to be 10mA?
Do you need to vary the current value and/or waveform shape?

Last edited: Jun 18, 2015
10. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0
I would say with less than 0.5 mA deviation.

10 mA is max current, but ideally I want mirroring resistor to be adjustable.
but current will stay constant if varied, constant 5 mA for example.

I will use square wave only, but I will change frequency and pulse width, I have developed a software that can produce different square waves.

11. ### crutschow Expert

Mar 14, 2008
13,000
3,229
Does the load have to be grounded?

Jun 5, 2015
30
0

13. ### kevin0228ca Thread Starter Member

Jun 5, 2015
30
0
actually I am able to get square wave signal with current mirror supplied by charge pump with my circuit, but signal is really noisy.
when I supplied with only 12V, signal was clear.

I also find that if I supply with C5, then signal is more clear than C7,
basically, C5>C7>C9>C11 signal clearness, and my current mirror is still constant voltage, so I think current mirror is working.
However, if I increase load from 1.5k to 2k+, Vsupply decreases instead of supplying maximum 40V, only supplied 28V.

anyone know if error is in current mirror or charge pump?

---------------------------------------------------------------------
I think problem is selection of 2n3906 or IRFH530, because I tested with Vsupply = C5 = about 20V, signal was clear.

If Vsupply = C7 = about 27V, signal is noisy.

Then I tested with 27V power supply instead of charge pump, signal disappear.

So I think my current mirror cannot handle Vsupply > 27V.

How can I choose correct pnp and mosfet?

Last edited: Jun 19, 2015
14. ### crutschow Expert

Mar 14, 2008
13,000
3,229
All the capacitor polarities need to be reversed in your schematic.
C4 and R5 are not needed.

It's much easier to design a constant current source if the load does not have to be grounded.
For example you can just put the load between V+ and the drain of your MOSFET, and eliminate the current-mirror.

15. ### Bordodynov Active Member

May 20, 2015
637
188
I can caution you from the errors of design of work of NE555. Many spice simulators in the model of timer show the overpriced output signal equal to VCC. On it the result of simulation will be too optimistic. I offer the variant. For strengthening of pumping signal I put an additional amplifaer. Requirements to working tensions of filter condensers are less hard. The values of capacity of these condensers must be anymore. It will decrease the pulsations of output tension.

16. ### DickCappels Moderator

Aug 21, 2008
2,651
632
A quick note: I don't know of any 2N3906 that is rated at 50 volts. You might want to change that before making a real-world circuit.

17. ### Bordodynov Active Member

May 20, 2015
637
188
See http://www.nxp.com/documents/data_sheet/PBSS4041SP.pdf
I think in this pair of transistors variation of tension Base-emitter will be less than, than at two separate transistors. But I nevertheless recommend in the emitters of transistors to put resistors 30 Ом. Most quality generators on four transistors.

• ###### Current.PNG
File size:
51.4 KB
Views:
2
Last edited: Jun 19, 2015