Hi,
is anyone having trouble getting the reg to start?
Without a load on the output (just the LED+r) I find I need 19.6Vdc on the AC input to get it to start. Vin=19V, Vout~=2V. Acceptable but high.
With a 100r load on the output neither the +ve nor -ve supplies will start, even with 31.6Vdc on each of the inputs.
I am using a dual lab supply with the current limit set to 200mA and can ramp up the voltage while monitoring either and/or both the D7 or D9 voltage and output voltage.
It appears to be the D7/9 voltage that refuses to ramp up forcing the regulator to hold the output low.
Would a Zener substituted for D1 and D16 allow start up on load?
Once started it regulates well, about .01% for input voltage variations and similar for zero to 150mA currents .
I am finding good LDO characteristics. 640mV drop (from D3 anode to output) on zero output and 860mV drop on 150mA output.
Bob,
how did you measure your 2.1V drop?
Once I have the self starting sorted with your help, next stage is checking for that 5MHz noise.
is anyone having trouble getting the reg to start?
Without a load on the output (just the LED+r) I find I need 19.6Vdc on the AC input to get it to start. Vin=19V, Vout~=2V. Acceptable but high.
With a 100r load on the output neither the +ve nor -ve supplies will start, even with 31.6Vdc on each of the inputs.
I am using a dual lab supply with the current limit set to 200mA and can ramp up the voltage while monitoring either and/or both the D7 or D9 voltage and output voltage.
It appears to be the D7/9 voltage that refuses to ramp up forcing the regulator to hold the output low.
Would a Zener substituted for D1 and D16 allow start up on load?
Once started it regulates well, about .01% for input voltage variations and similar for zero to 150mA currents .
I am finding good LDO characteristics. 640mV drop (from D3 anode to output) on zero output and 860mV drop on 150mA output.
Bob,
how did you measure your 2.1V drop?
Once I have the self starting sorted with your help, next stage is checking for that 5MHz noise.
The only time I have had trouble with the regulator not starting was with very low resistance loads 25 ohms or so.
D1 and D16 prevent damage to the regulator in the event that the input voltage drops below the output voltage on shut down. They should stay as designed.
R3 and R4 form a bit of short circuit protection. If the load is low enough, A voltage divider formed by R3 and R4 keeps the base of T3 at a low voltage to prevent it conducting, therefore the pass devices cannot turn on. This saved me a couple times on the prototype boards without a soldermask.
D6 and R7 also affect turn on behavior through a mechanism I hope Jens will explain. However, it seems that if D6 were installed backwards, it might affect startup reliability. I think they prevent overshoot.
Andrew, I hope that I do not offend by suggesting that you verify the values of R3, R4, R7 and orientation of D6. (and the equivalent parts on the negative side). Try lifting one end of R4 to see if it starts reliably - you will need to so you can measure it anyway. Performance may not be as good, and you'll lose any short circuit protection, so be careful.
The 2.1V I measured as the dropout was measured quite crudely - It was what I saw on my voltmeter when ripple began to appear on the output on my scope. If you are measuring from the bottom of the output ripple to output voltage, you have a more accurate measure of dropout performance.
I am down to one working scope lead, so I cannot be more accurate for the time being. (I know, whip all abusers of oscilloscope leads)
D1 and D16 prevent damage to the regulator in the event that the input voltage drops below the output voltage on shut down. They should stay as designed.
R3 and R4 form a bit of short circuit protection. If the load is low enough, A voltage divider formed by R3 and R4 keeps the base of T3 at a low voltage to prevent it conducting, therefore the pass devices cannot turn on. This saved me a couple times on the prototype boards without a soldermask.
D6 and R7 also affect turn on behavior through a mechanism I hope Jens will explain. However, it seems that if D6 were installed backwards, it might affect startup reliability. I think they prevent overshoot.
Andrew, I hope that I do not offend by suggesting that you verify the values of R3, R4, R7 and orientation of D6. (and the equivalent parts on the negative side). Try lifting one end of R4 to see if it starts reliably - you will need to so you can measure it anyway. Performance may not be as good, and you'll lose any short circuit protection, so be careful.
The 2.1V I measured as the dropout was measured quite crudely - It was what I saw on my voltmeter when ripple began to appear on the output on my scope. If you are measuring from the bottom of the output ripple to output voltage, you have a more accurate measure of dropout performance.
I am down to one working scope lead, so I cannot be more accurate for the time being. (I know, whip all abusers of oscilloscope leads)
Hi Bob,
I am going to remove R4 from circuit. I expect it to start properly then. However that completely defeats the Vref fed from the regulated output, so having proved that it self starts I shall reinsert R4.
But are there any disadvantages in doing this? or is there a better way?
Jens or anyone,
can you explain the purpose of D6 & R7?
I am going to remove R4 from circuit. I expect it to start properly then. However that completely defeats the Vref fed from the regulated output, so having proved that it self starts I shall reinsert R4.
Agreed, but a zener in here achieves the same reverse voltage protection upto the limits of the zener in forward conduction mode. In reverse mode when a high voltage appears across the pass transistor the Zener would bypass T1, charge up the output and thereby feed the Vref as well as the load.
But are there any disadvantages in doing this? or is there a better way?
Jens or anyone,
can you explain the purpose of D6 & R7?
Hi all,
I snipped out R4 to allow Vref to to be fed from the input. As expected the PSU starts properly with a 100r load connected.
The Vref voltage tracks the input voltage at just less than 50% of Vin. At about Vin=11V the Vref has reached 4.7V and suddenly switches to 8.8V as the output comes on stream at 7V. It appears that D6 + R7 hold Vref at average of Voutput & Vin and when T1 switches on bringing up Voutput then Vref follows. Voutput then tracks Vin until regulation is achieved (in my case 15V).
This all seems quite sensible.
How do we achieve a similarly controlled switch on when Vref is fed from the output? Is there a design flaw in the switch on mode? Can it be overcome with this PCB?
I would like to remove R3 to feed a clean supply to Vref. R3 appears to have been put in to try to achieve a switch on mode, but the ratio of R3 to R4//R7 holds Vref too low during start up. It is a poor switch on mode that inherently fails to achieve it's purpose.
There must be a better way. Or have I misunderstood the mechanisms at work here?
I have just wound up Vout to 29V and Vin @ 31.6V (before the diodes). Vdrop at Iout=290mA is still only 820mV. Superb. T1 is only warm (Rths-a=21C/W). Rload (10W 100r) is smelling a bit now.
I snipped out R4 to allow Vref to to be fed from the input. As expected the PSU starts properly with a 100r load connected.
The Vref voltage tracks the input voltage at just less than 50% of Vin. At about Vin=11V the Vref has reached 4.7V and suddenly switches to 8.8V as the output comes on stream at 7V. It appears that D6 + R7 hold Vref at average of Voutput & Vin and when T1 switches on bringing up Voutput then Vref follows. Voutput then tracks Vin until regulation is achieved (in my case 15V).
This all seems quite sensible.
How do we achieve a similarly controlled switch on when Vref is fed from the output? Is there a design flaw in the switch on mode? Can it be overcome with this PCB?
I would like to remove R3 to feed a clean supply to Vref. R3 appears to have been put in to try to achieve a switch on mode, but the ratio of R3 to R4//R7 holds Vref too low during start up. It is a poor switch on mode that inherently fails to achieve it's purpose.
There must be a better way. Or have I misunderstood the mechanisms at work here?
I have just wound up Vout to 29V and Vin @ 31.6V (before the diodes). Vdrop at Iout=290mA is still only 820mV. Superb. T1 is only warm (Rths-a=21C/W). Rload (10W 100r) is smelling a bit now.
I wouldn't say that the choice of R3 "inherently fails to achieve its purpose," as you are the only one reporting startup difficulties. I appreciate your frustration, but the design appears to work as is. I've got four of these working, Jens has at least two and and there are a couple others reported here. Please verify that your component values are correct and do the usual solder bridge/bad joint check.
I'd still check resistor values - including R10 (is there enough current available in the differential to start T1) and R1 (is that current developing enough voltage to turn on T1, T2)
How about R9? If too high it could take a while to charge C15 and allow T3 to turn on.
Just to try to cover all bases, is your load resistor really 100 ohms, not 10?
Can you build up another of your boards and see if you have the same issues?
A SIDE NOTE for those just looking for a plug and play regulator - a 15VAC transformer will give around 20V after the rectifiers, enough to start even Andrew's stubborn regulator. If he feeds 19.6 VDC into the AC terminals, he's getting 18.4V or so at the regulator input due to the rectifier voltage drops.
I'd still check resistor values - including R10 (is there enough current available in the differential to start T1) and R1 (is that current developing enough voltage to turn on T1, T2)
How about R9? If too high it could take a while to charge C15 and allow T3 to turn on.
Just to try to cover all bases, is your load resistor really 100 ohms, not 10?
Can you build up another of your boards and see if you have the same issues?
A SIDE NOTE for those just looking for a plug and play regulator - a 15VAC transformer will give around 20V after the rectifiers, enough to start even Andrew's stubborn regulator. If he feeds 19.6 VDC into the AC terminals, he's getting 18.4V or so at the regulator input due to the rectifier voltage drops.
Hi,
I have checked all my components, yes, some resistors had been changed to values I thought better. These all bar one were returned to the published values before I retested and posted my first enquiry (post321). I still have R10 = 3k6 and plan to change that one now.
I found that the circuits, both +ve & -ve would reluctantly start without a load but with a 100r load would not start on Vin=31.6V. Now that R4 is removed both start perfectly and both regulate correctly with or without R4.
Both regulate correctly when using my old resistor values.
Most of my caps are not as published, ceramic substituted for 1uFpes due to shortage of stock.
I have described the start up sequence and the initial lack of voltage across Vref. This appears to be the source of the start up problem.
The voltage divider formed by R3 + R4//R7 allows only 8.3% of the difference between Vout and Vin to appear at Vref.
With Vin=30V and Vout=0.1V then Vref will be about 2.5V. That's why I suggested that R3 inherently fails to allow start up. If that is it's purpose. Co-incidentally, I asked R3's purpose some months ago and got no answer.
I ask again, have I understood the operational mechanisms at work during start up?
Could a prod from the oscillations of the transformer/rectifiers be enough to start up a fully working version?
I ask, since an easy mod would be to pre-regulate using a floating LM317/337, but this would eliminate transformer variations from our regulator.
I have checked all my components, yes, some resistors had been changed to values I thought better. These all bar one were returned to the published values before I retested and posted my first enquiry (post321). I still have R10 = 3k6 and plan to change that one now.
I found that the circuits, both +ve & -ve would reluctantly start without a load but with a 100r load would not start on Vin=31.6V. Now that R4 is removed both start perfectly and both regulate correctly with or without R4.
Both regulate correctly when using my old resistor values.
Most of my caps are not as published, ceramic substituted for 1uFpes due to shortage of stock.
I have described the start up sequence and the initial lack of voltage across Vref. This appears to be the source of the start up problem.
The voltage divider formed by R3 + R4//R7 allows only 8.3% of the difference between Vout and Vin to appear at Vref.
With Vin=30V and Vout=0.1V then Vref will be about 2.5V. That's why I suggested that R3 inherently fails to allow start up. If that is it's purpose. Co-incidentally, I asked R3's purpose some months ago and got no answer.
I ask again, have I understood the operational mechanisms at work during start up?
Could a prod from the oscillations of the transformer/rectifiers be enough to start up a fully working version?
I ask, since an easy mod would be to pre-regulate using a floating LM317/337, but this would eliminate transformer variations from our regulator.
Hi,
r10 = 1k06 and R4 =1k05 re-inserted.
Startup characteristics unchanged.
Rload=100r, Vin=31.7V, Vref = 2.1V, Vout = 0.24V
Disconnect Rload and regulation back to normal, reconnect Rload and Vout drops to 14.99V, excellent.
But the lack of Vref still stops it starting "on load" which should be the normal method of operation.
What has become apparent is that regulation accuracy is very tolerant of resistor value and capacitor type changes. That indicates that the general circuit topology is not dependant on accurate component selection. Another good sign.
r10 = 1k06 and R4 =1k05 re-inserted.
Startup characteristics unchanged.
Rload=100r, Vin=31.7V, Vref = 2.1V, Vout = 0.24V
Disconnect Rload and regulation back to normal, reconnect Rload and Vout drops to 14.99V, excellent.
But the lack of Vref still stops it starting "on load" which should be the normal method of operation.
What has become apparent is that regulation accuracy is very tolerant of resistor value and capacitor type changes. That indicates that the general circuit topology is not dependant on accurate component selection. Another good sign.
Hi guys,
I built one of the regs a couple of weeks ago and it seems to be working fine, I've been using it to power a JLH chiarra classA headphone amp for testing, the heatsinks get very warm after an hours use but the voltage out seems nice and stable🙂
I'm using 1n5819 schottkie diodes
http://pic20.picturetrail.com/VOL1293/4634025/9699445/155245036.jpg
I built one of the regs a couple of weeks ago and it seems to be working fine, I've been using it to power a JLH chiarra classA headphone amp for testing, the heatsinks get very warm after an hours use but the voltage out seems nice and stable🙂
I'm using 1n5819 schottkie diodes
http://pic20.picturetrail.com/VOL1293/4634025/9699445/155245036.jpg
An externally hosted image should be here but it was not working when we last tested it.
R3 is intended to be a ”boot” resistor value. The ratio of 1:10 between R4 and R3 should suppress the supply ripple enough to give a decent reference voltage.
R7 helps to shut down the regulator under a short circuit situation by pulling the ref voltage down. In a fault where Vout is below Vref it is in parallel with R4 and shunts the ref voltage. If you R7 value is too small it will not allow the regulator to start as vref is pulled down to the 0V output voltage. Please verify the R7 is the specified value.
I’m not convinced that the optimal R7 value is 10k, but it has worked in my test setups. Also the value of R4 may not be optimal; it might be better in the 10k range or similar to what is used for R7
Anyway, please report back what you find and let us adjust the BOM values as we go along….
Thanks for your hard work Andrew.
\Jens
R7 helps to shut down the regulator under a short circuit situation by pulling the ref voltage down. In a fault where Vout is below Vref it is in parallel with R4 and shunts the ref voltage. If you R7 value is too small it will not allow the regulator to start as vref is pulled down to the 0V output voltage. Please verify the R7 is the specified value.
I’m not convinced that the optimal R7 value is 10k, but it has worked in my test setups. Also the value of R4 may not be optimal; it might be better in the 10k range or similar to what is used for R7
Anyway, please report back what you find and let us adjust the BOM values as we go along….
Thanks for your hard work Andrew.
\Jens
Thanks Jens,
gives me more understanding of what is supposed to happen.
I am now inclined to argue that the 1:10 ratio R3:R4 is not optimal for the start up mode. But the consequence of reducing this ratio moves the regulator further away from Vref fed from Vout.
Is there another way to ensure reliable start up, preferably without R3. Would a Diac for R3 be a solution?
gives me more understanding of what is supposed to happen.
I am now inclined to argue that the 1:10 ratio R3:R4 is not optimal for the start up mode. But the consequence of reducing this ratio moves the regulator further away from Vref fed from Vout.
Is there another way to ensure reliable start up, preferably without R3. Would a Diac for R3 be a solution?
I have no problems with starting the regulator at all.Tried with or without a 150R load and ac voltage as low as 14.2V.
Are you shure your current limiting at start-up is not the
fault.Charging the caps for the first time will normally require more allthough very briefly.Limiting this start-up current will result in a slower rise time of the input voltage.
Kees.
Are you shure your current limiting at start-up is not the
fault.Charging the caps for the first time will normally require more allthough very briefly.Limiting this start-up current will result in a slower rise time of the input voltage.
Kees.
To make shure i rigged up a variac in front of the toroid.
Without load,just the LED i got the following.
No output until input voltage is 5.24Vac.
At 5.24Vac in i get 5.10Vdc out.
Increasing the ac also increases the dc until the ac gets to 13.78Vac then i get 15.00Vdc out.Further increasing the ac has no effect on the output.
With a 150R load these values change.
No output until input voltage is 12.58Vac.
At 12.58Vac in i get 12.89Vdc out.
At 14.25Vac in i get 15.00Vdc out.
These values remain the same after several tries.
Without load,just the LED i got the following.
No output until input voltage is 5.24Vac.
At 5.24Vac in i get 5.10Vdc out.
Increasing the ac also increases the dc until the ac gets to 13.78Vac then i get 15.00Vdc out.Further increasing the ac has no effect on the output.
With a 150R load these values change.
No output until input voltage is 12.58Vac.
At 12.58Vac in i get 12.89Vdc out.
At 14.25Vac in i get 15.00Vdc out.
These values remain the same after several tries.
Good point on the current limiting Kees. Andrew, are you able to increase the limit current on your supply?
Hi,
it's worth a try. Last job before I get off to bed.
Kro's number look interesting.
First look suggests that no load switch on occurs at low AC input (5.24Vac).
At Rload=150r the circuit needs 2.5 times the AC input to switch on (12.58Vac).
After switch on the output tracks the input until regulated Vout is reached. Similar symptoms.
it's worth a try. Last job before I get off to bed.
Kro's number look interesting.
First look suggests that no load switch on occurs at low AC input (5.24Vac).
At Rload=150r the circuit needs 2.5 times the AC input to switch on (12.58Vac).
After switch on the output tracks the input until regulated Vout is reached. Similar symptoms.
Hi,
that didn't help.
Set current limit to 1A and connected input straight to live PSU output. Current limiting light flashed momentarily then Vin reset to 31.7Vdc and output stayed resolutely at 0.24Vdc.
Removed Rload reset Vin to 22Vdc and made input connection. Again current limit light flickered but Vout came almost straight up to 15Vdc. So same problem, on load start up still not working.
that didn't help.
Set current limit to 1A and connected input straight to live PSU output. Current limiting light flashed momentarily then Vin reset to 31.7Vdc and output stayed resolutely at 0.24Vdc.
Removed Rload reset Vin to 22Vdc and made input connection. Again current limit light flickered but Vout came almost straight up to 15Vdc. So same problem, on load start up still not working.
After you get some rest, Would you check Vref and the voltages across R10 and R1 in the "refuse to start mode"? (did you change R10 back to 1K?) I'm curious what the total current is in the differential and the current in the T3 leg.
Something makes me think you may have a bad T3...
Something makes me think you may have a bad T3...
Hi Bob,
If the regulation works well, both in rejecting input voltage variations and relative immunity from output current variations, then surely the transistors must be OK.
At the moment my DMM is working at resolution limit to show 14.98V to 14.99V minimum and 15.00V to 15.01V maximum under all conditions.
I do have a 5.5 digit DMM but have not brought it into regular use. Seems like today might be the opportunity to check whether it's any good.
Thanks for your continued support.
Vr1 & Vr10 will have to wait till I get home from work.
If the regulation works well, both in rejecting input voltage variations and relative immunity from output current variations, then surely the transistors must be OK.
At the moment my DMM is working at resolution limit to show 14.98V to 14.99V minimum and 15.00V to 15.01V maximum under all conditions.
I do have a 5.5 digit DMM but have not brought it into regular use. Seems like today might be the opportunity to check whether it's any good.
Thanks for your continued support.
Yeah, I was a little brain dead when I suggested T3 was bad.
I still wonder if the current limiting in your supply may be causing a latch-up in the non operational condition, since you get current limiting indicator blinks.
Can you feed your supply 15 VAC and see if it starts under more normal operating conditions?
I still wonder if the current limiting in your supply may be causing a latch-up in the non operational condition, since you get current limiting indicator blinks.
Can you feed your supply 15 VAC and see if it starts under more normal operating conditions?
Hi,
success, well almost.
Bob,
thanks for your persistence.
Reducing r10 from 3k6 to 1k06, I paralleled R7 by mistake, PCB upside down, we could bet an antipodean wouldn't have done that.
Now that R10 really is 1k06 it switches on at 19.9Vdc with 100R load (150mA output) on line.
Now for your numbers
Vin=19.8Vdc, Vout=0.22V, Vr1=0.9V, Vr10=0.93V.
Vin=19.9Vdc, Vout=15.06V, Vr1=1.28V, Vr10=8.9V
off state: Ir1~=Ir10=0.9mA, indicating that T4 is off.
on state: Ir1=8.9mA, Ir10=1.3mA, indicating T4 is passing 7.6mA.
When used as a high voltage regulator (maybe 40Vout to 60Vout), T4 is going to run mighty hot! Which is part of the reason I increased R10 (to 3k6) to balance the LTP currents and reduce dissipation on the inverting leg. But it now transpires this is at odds with online start up.
On the same tack, using this regulator to power KSA & Leach front ends, I would be expecting to run pre-smoothed supplies from the main PSU with a piggy backed LV winding to give the pre regulator voltage. This type of start up will more closely mimic the lab supply than a high regulation dedicated transformer that will show a high peak voltage prior to the regulators starting up.
The other problem is the inability to use a floating pre-regulator set to 2.3V above Vout. It just won't start.
So, although I have a solution, it is only partial success.
Back to my initial request.
Is there a way to add or modify some components to achieve a better start up mode?
success, well almost.
Bob,
thanks for your persistence.
Reducing r10 from 3k6 to 1k06, I paralleled R7 by mistake, PCB upside down, we could bet an antipodean wouldn't have done that.
Now that R10 really is 1k06 it switches on at 19.9Vdc with 100R load (150mA output) on line.
Now for your numbers
Vin=19.8Vdc, Vout=0.22V, Vr1=0.9V, Vr10=0.93V.
Vin=19.9Vdc, Vout=15.06V, Vr1=1.28V, Vr10=8.9V
off state: Ir1~=Ir10=0.9mA, indicating that T4 is off.
on state: Ir1=8.9mA, Ir10=1.3mA, indicating T4 is passing 7.6mA.
When used as a high voltage regulator (maybe 40Vout to 60Vout), T4 is going to run mighty hot! Which is part of the reason I increased R10 (to 3k6) to balance the LTP currents and reduce dissipation on the inverting leg. But it now transpires this is at odds with online start up.
On the same tack, using this regulator to power KSA & Leach front ends, I would be expecting to run pre-smoothed supplies from the main PSU with a piggy backed LV winding to give the pre regulator voltage. This type of start up will more closely mimic the lab supply than a high regulation dedicated transformer that will show a high peak voltage prior to the regulators starting up.
The other problem is the inability to use a floating pre-regulator set to 2.3V above Vout. It just won't start.
So, although I have a solution, it is only partial success.
Back to my initial request.
Is there a way to add or modify some components to achieve a better start up mode?
- Status
- Not open for further replies.