Andrew T said:
from my earlier report:
The CCS controlling current through Vref is very sensitive to temperature. I have now got output voltage holding consistently to better than 0.05%/Vinput. Yes <5mVdc change in both a 5V and an 8.4V reg for a 10V change in the DC supply voltage.
But the temperature coefficients are all over the place, just as I said earlier.
The worst is the CCS jFET and this should be a low end of BL grade to minimise volt drift vs temp. But we still need a good Vref that varies little with changes in temperature and with changes in current. A resistor Vref cannot do this. 1% change in current = 1% change in voltage. or put another way 5mV change in voltage will result from a 0.1% change in the CCS current. Resistor Vref is asking for impossible drift performance with our reg.
LED Zener and Diode all vary in voltage vs temperature. But they all perform far better in respect of voltage variation vs CCS current change. I believe that for low Vdrift the Vref must use the minimum of resistor value and maximise the use of passive voltage regulators.
Last edited:
You are right Andrewt, for low voltage will be better the Vref with more LEDs instead 5K trimmer & to use a low ppm fixed resistor to adjust Vout, all this info repeat several time Salas on his thread.
Wich voltage are the LEDs are you using? If your target Vout are 15V instead VR1 you can use several LEDs + diodes + final low ppm fixed resistor to got 15Vout
hi Andrew T
would you draw a simple schematic so I can better follow what to change,your better in touch with this than I am,Would a adjustable lm317 driving a fet get us there or is that regressing?
Thanks
would you draw a simple schematic so I can better follow what to change,your better in touch with this than I am,Would a adjustable lm317 driving a fet get us there or is that regressing?
Thanks
I suggest the maximum value of VR trimmer + R fixed resistor should be less than 200r.
+-1% CCS current variation of an 8mA Idss jFET passing 7mA through the Vref gives a +-14mV variation in Vref from the resistance alone.then you have to add on the temp co of the two resistances.
+-1% current variation for the jFET CCS that has just 650mV to operate with is not at all easy for all temperature variations that will occur.
Using 1k or 5k or 10k for VR in a Vref is getting silly.
eg. Vr=100r and set to 20r to trim output voltage.
+-2% on 7mA CCS gives ~+-3mV on output voltage. The high temp co VR can then be replaced with a low temp co fixed resistor.
LEDS (red orange yellow and green are all suitable) signal diodes can make up low voltage Vref. The Quanghao board has plenty space for 4 devices with minor trace cutting. This would suit voltage outputs <9Vdc.
For higher output voltages one LED/diode can be replaced with a Zener//cap (to reduce noise slightly).
I have moved the trim to become the jFET source resistor. This removes the resistor variable voltage effect from the Vref and can be fitted into Quanghao board.
+-1% CCS current variation of an 8mA Idss jFET passing 7mA through the Vref gives a +-14mV variation in Vref from the resistance alone.then you have to add on the temp co of the two resistances.
+-1% current variation for the jFET CCS that has just 650mV to operate with is not at all easy for all temperature variations that will occur.
Using 1k or 5k or 10k for VR in a Vref is getting silly.
eg. Vr=100r and set to 20r to trim output voltage.
+-2% on 7mA CCS gives ~+-3mV on output voltage. The high temp co VR can then be replaced with a low temp co fixed resistor.
LEDS (red orange yellow and green are all suitable) signal diodes can make up low voltage Vref. The Quanghao board has plenty space for 4 devices with minor trace cutting. This would suit voltage outputs <9Vdc.
For higher output voltages one LED/diode can be replaced with a Zener//cap (to reduce noise slightly).
I have moved the trim to become the jFET source resistor. This removes the resistor variable voltage effect from the Vref and can be fitted into Quanghao board.
Last edited:
hi Andrew T
So could you do a layout and show what to change ?
Have you done this mod and does it indeed work ,Temp fluctuation better?
So could you do a layout and show what to change ?
Have you done this mod and does it indeed work ,Temp fluctuation better?
Yes, it works. All my previous are reports on my experiments, they are not a work of fiction.
No, none of my modifications reduce temperature fluctuations.
No modification to the regulator can alter or reduce the changes in ambient temperature.
None of the changes that I have proposed can alter the variation in power dissipation as mains voltage changes.
Hi Andrew T,
looks like you done as much tweeting as it can get,
Have you thought about sharing what you think would be a ultimate design,
a schematic or layout ,I always like to see a new idea come to frutation,
I will have time to correct my boards from my parts errors I did tomorrow,and I'll run them to see how they do,,,,,,,,,
looks like you done as much tweeting as it can get,
Have you thought about sharing what you think would be a ultimate design,
a schematic or layout ,I always like to see a new idea come to frutation,
I will have time to correct my boards from my parts errors I did tomorrow,and I'll run them to see how they do,,,,,,,,,
yes, but I have others in the queue for my time before I can do a layout.
I can scribble on a print and photograph the trace cuts and the experimental version of the final Vref string and post these. Probably later today.
Yes, this really does need a new layout and that was why I asked quite a while back "what do you need?". I got one or two replies.
I can scribble on a print and photograph the trace cuts and the experimental version of the final Vref string and post these. Probably later today.
Yes, this really does need a new layout and that was why I asked quite a while back "what do you need?". I got one or two replies.
ooops but now it works....
Hi all,
Merlin was correct ,I read some more and had forgot it needed a 5 k trimmer( Parts almost took 2 months to get here ),that part is working putting 40 ac in 43 vdc ar R1 and got 38.6 at the out puts,not loaded,I loaded them with a 150 r 10 watt and set it on 30 volts on outputs and watched,,,,,,after about 5 mins it had moved .04V down so 29.96 and stayed put, the heatsinks, (ambient temp was 70) the middle heatsink was at 86 F and the other at 76 F,all running good so far, now I have change the other board to test tonight,I change the heat sinks, now I have the large one like in the picture a few post back in the middle and a smaller about half the size at the output,It will run tonight if all goes well into service for a NoSMOKING test,lol.
A question for anyone ???does the heatsink heat varry as the signal demands or is there such a time lag that it is not a factor ,large caps take care of this with smaller caps to get what's left, IS this why a bigger heatsink is better? truly a stabilizer ,
Thanks for all the help!!!!!
Hi all,
Merlin was correct ,I read some more and had forgot it needed a 5 k trimmer( Parts almost took 2 months to get here ),that part is working putting 40 ac in 43 vdc ar R1 and got 38.6 at the out puts,not loaded,I loaded them with a 150 r 10 watt and set it on 30 volts on outputs and watched,,,,,,after about 5 mins it had moved .04V down so 29.96 and stayed put, the heatsinks, (ambient temp was 70) the middle heatsink was at 86 F and the other at 76 F,all running good so far, now I have change the other board to test tonight,I change the heat sinks, now I have the large one like in the picture a few post back in the middle and a smaller about half the size at the output,It will run tonight if all goes well into service for a NoSMOKING test,lol.
A question for anyone ???does the heatsink heat varry as the signal demands or is there such a time lag that it is not a factor ,large caps take care of this with smaller caps to get what's left, IS this why a bigger heatsink is better? truly a stabilizer ,
Thanks for all the help!!!!!
Last edited:
Hi,
the regulator has 2 main parts, the CCS and the Shunt reg.
As temperature varies the Vgs vs Id varies.
The CCS responds by changing the CCS value slightly. As mains input voltage varies the power dissipated by the CCS varies. This in turn changes temperatures inside the case and inside the CCS FET, so again CCS current varies slightly . This change in CCS current has virtually no effect on output voltage. The power dissipated has a profound effect on the Vout due to what is explained below.
The Vgs vs Id variation with temperature of the Shunt device has very little effect on output voltage.
The output voltage is determined by that formula quoted earlier.
Vout = Vref + Vdropstopper + Vbecontrol
Vref and Vbecontrol are affected by temperature.
An increase in output current demand will reduce the dissipation of the Shunt device. This cools the regulator. A reduction in mains voltage will cool the regulator. The tempco of the various temperature sensitive devices will change the output voltage. Some tempco are +ve, some are -ve, many are quite small. One tempco is large if the wrong jFET is selected. This can swamp all other Vout variations.
the regulator has 2 main parts, the CCS and the Shunt reg.
As temperature varies the Vgs vs Id varies.
The CCS responds by changing the CCS value slightly. As mains input voltage varies the power dissipated by the CCS varies. This in turn changes temperatures inside the case and inside the CCS FET, so again CCS current varies slightly . This change in CCS current has virtually no effect on output voltage. The power dissipated has a profound effect on the Vout due to what is explained below.
The Vgs vs Id variation with temperature of the Shunt device has very little effect on output voltage.
The output voltage is determined by that formula quoted earlier.
Vout = Vref + Vdropstopper + Vbecontrol
Vref and Vbecontrol are affected by temperature.
An increase in output current demand will reduce the dissipation of the Shunt device. This cools the regulator. A reduction in mains voltage will cool the regulator. The tempco of the various temperature sensitive devices will change the output voltage. Some tempco are +ve, some are -ve, many are quite small. One tempco is large if the wrong jFET is selected. This can swamp all other Vout variations.
Last edited:
Connect two wires for +V and two wires for GND to your load. Get them back to the reg board to the corresponding out and sense pairs.
DON'T jumper the little white lines on board next to the sense marks. As why I did it, read this: Kelvin (4-wire) resistance measurement : DC METERING CIRCUITS Will help keep reg's Zo as it should. Brings reg and load ''closer''.
What you may care to know is it will benefit your sonics also. Trust me on that and use it.
DON'T jumper the little white lines on board next to the sense marks. As why I did it, read this: Kelvin (4-wire) resistance measurement : DC METERING CIRCUITS Will help keep reg's Zo as it should. Brings reg and load ''closer''.
What you may care to know is it will benefit your sonics also. Trust me on that and use it.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.