Hello,
I have these regulators as part of a circuit: Schematics.gif
I have a problem: The measured output is -3.74/+3.74v instead of -5/+5v. I cahnged the 5.6k resistor for a 2.4k, the output changed to 4 volts. I changed the 2.4k resistors at the output of the opamps to 1.2k with no change in output voltage. Measured voltage drop on zeners at 3.74v (same as output).
Should I increase the Zener diodes value or decrease the 5.6k resistor in order to achieve the desired -5/+5v output?
About the 2sk246 FET: is the Source and Drain reversible? I wonder if the schematics reversed the two. Whould it make a difference?
Thanks,
Mihai
I have these regulators as part of a circuit: Schematics.gif
I have a problem: The measured output is -3.74/+3.74v instead of -5/+5v. I cahnged the 5.6k resistor for a 2.4k, the output changed to 4 volts. I changed the 2.4k resistors at the output of the opamps to 1.2k with no change in output voltage. Measured voltage drop on zeners at 3.74v (same as output).
Should I increase the Zener diodes value or decrease the 5.6k resistor in order to achieve the desired -5/+5v output?
About the 2sk246 FET: is the Source and Drain reversible? I wonder if the schematics reversed the two. Whould it make a difference?
Thanks,
Mihai
No the schematic seems ok.
The zener must be faulty if the voltage drop isn't more than 4v when you have the 2.4k resistor.
The current through the resistor and zener should be ~1.6mA which should be enough.
Change the zener, or check its datasheet to see at which current it has 5v.
Edit: The fet might be wrong but i am not sure.
The zener must be faulty if the voltage drop isn't more than 4v when you have the 2.4k resistor.
The current through the resistor and zener should be ~1.6mA which should be enough.
Change the zener, or check its datasheet to see at which current it has 5v.
Edit: The fet might be wrong but i am not sure.
I think also the design is a bit "twisted".
Jan, it will probably work but the "twisted" thing is that the negative voltage will have influence to the positive.
First you must calculate the voltages you should have and of course check all parts, the inputs of the opamps, are the correctly connected?
Notice also that a 5.1 V(?) zener have tolerances and is also not so "sharp". 5.6 kohms = only 1 mA, which is rather little. I had chosen 2-5 mA at least.
If you are unsure about the zeners, jsut test them only with a resistor and the zener. Then you can see what can expect in voltage.
Mihai, check also if the voltage between the inputs of the opams are zero (less than 1-4 mV) even when you load the regulator. If yes, it seems to work. The regulator may also oscillate. Have you checked the output with an oscilloscope?
How does you pcb look like? You may have problems there?
Jan, it will probably work but the "twisted" thing is that the negative voltage will have influence to the positive.
First you must calculate the voltages you should have and of course check all parts, the inputs of the opamps, are the correctly connected?
Notice also that a 5.1 V(?) zener have tolerances and is also not so "sharp". 5.6 kohms = only 1 mA, which is rather little. I had chosen 2-5 mA at least.
If you are unsure about the zeners, jsut test them only with a resistor and the zener. Then you can see what can expect in voltage.
Mihai, check also if the voltage between the inputs of the opams are zero (less than 1-4 mV) even when you load the regulator. If yes, it seems to work. The regulator may also oscillate. Have you checked the output with an oscilloscope?
How does you pcb look like? You may have problems there?
Thank you all for the suggestions.
The design is from china, probably a copy of something else?
The pcb is ok, no errors there. I have 4 of these working, one which is using 15v Zeners. It's output is at 14v (acceptable). The other 3 are 5v and all behave the same way. The chinese schematics mention using a op275 as and option instead of the op249. I don't think it will change much.
I'll try increasing the current on the zeners as you suggested.
The design is from china, probably a copy of something else?
The pcb is ok, no errors there. I have 4 of these working, one which is using 15v Zeners. It's output is at 14v (acceptable). The other 3 are 5v and all behave the same way. The chinese schematics mention using a op275 as and option instead of the op249. I don't think it will change much.
I'll try increasing the current on the zeners as you suggested.
How would you give current to the zeners if you reference them to ground? You would have to tap it somewhere before the transistors. You would also need to put the sensing on the vout?
It does have a cap across the zener but admittedly it is not that big. Guesstimating you have at least 45dB damping from the other rail ripple.
Yes i think the regulator is very load sensitive but its PSRR should be high. Maby it is intended for class A use.
It does have a cap across the zener but admittedly it is not that big. Guesstimating you have at least 45dB damping from the other rail ripple.
Yes i think the regulator is very load sensitive but its PSRR should be high. Maby it is intended for class A use.
hjelm said:
You want to regulate the output voltage referenced to ground. So you develop a reference voltage to ground for the + input of the error amp. You do this by a resistor from + (or -) to a zener to ground, and feed the zener voltage to the + input. Then you feed the sample of the output voltage to the - input. The error amp will do whatever is necessary to make its inputs equal. By adjusting the division of the output voltage before feeding it to the - input you set the output voltage. The reference at the + input should be as constant and stable as possibel, after all it is a 'reference'. So, you may want to include an RC filter between the zener and the + input. But that's details.
Jan Didden
But it is referenced to ground but the other way around. Is there a big problem with it? Either it ripples with the Supply or it ripples with the output. You could argue that the output is more stable than the supply? I can see the pont in having the supply react symmetrically if you use a differential architecture of the DAC but this circuit doesn't react symmetrically.
Not arguing here really but i just thought there might be a point to it.
Nobody has answered if the FET's are turned the right way around. Are they there as a CCS?
Not arguing here really but i just thought there might be a point to it.
Nobody has answered if the FET's are turned the right way around. Are they there as a CCS?
It seems to me that they made this strange circuit because of the need to supply the opamp from the +/- supply. The opamp inputs will be at ground potential (+ hardwired, the - because of the design). So, if the pos supply output goes up, the - input goes up, and the opamp regulates the output down. So far, so good, so it looks that it can indeed work - I should have looked at it a bit longer maybe. The output voltages are equal to the zener voltages - if everything works as designed.
The supplies DO influence each other, but that influence is damped by the ratio of the zener dynamic impedance to the 5.6 k resistors (which are a bit high, 1mA through the zeners is a bit low).
If it doesn't work, check that the opamp inputs are both at or close to (few mV) ground.
One check would be to replace the current source by a resistor of, say, 2 or 3 k (and decrease the 2.4k to 100 ohms or so), that should work also. The 2 or 3 k will insert too much current into the pass transistor base, which will be syphoned off by the opamp.
Jan Didden
The supplies DO influence each other, but that influence is damped by the ratio of the zener dynamic impedance to the 5.6 k resistors (which are a bit high, 1mA through the zeners is a bit low).
If it doesn't work, check that the opamp inputs are both at or close to (few mV) ground.
One check would be to replace the current source by a resistor of, say, 2 or 3 k (and decrease the 2.4k to 100 ohms or so), that should work also. The 2 or 3 k will insert too much current into the pass transistor base, which will be syphoned off by the opamp.
Jan Didden
I'll measure the opamp inputs later today and post the results.
I wonder about the FET because the PCB is marked reversed of the schematics: the Drain and Gate are connected instead of the Source/Gate. I wonder which is right? Presently I followed the schematics. I was thinking of reversing them, but I don't want to blow the transistors since they are hard to find here and it took me 2 weeks to get them.
Mihai
I wonder about the FET because the PCB is marked reversed of the schematics: the Drain and Gate are connected instead of the Source/Gate. I wonder which is right? Presently I followed the schematics. I was thinking of reversing them, but I don't want to blow the transistors since they are hard to find here and it took me 2 weeks to get them.
Mihai
Fred,
It does work I think. It looks like a clever way to cope with the limited input range of the opamp; here both inputs are at ground level, which of course is nicely in the middle of the supply range. The opamp output series R does the same to keep the opamp output inside the supply envelope. The other polarity is used to bias the zeners, so you need no output attenuator resistors to set the output level; the zener voltage IS the output voltage. It's creative, to say the least. Took me some time to look through it. Do you see it?
Jan Didden
It does work I think. It looks like a clever way to cope with the limited input range of the opamp; here both inputs are at ground level, which of course is nicely in the middle of the supply range. The opamp output series R does the same to keep the opamp output inside the supply envelope. The other polarity is used to bias the zeners, so you need no output attenuator resistors to set the output level; the zener voltage IS the output voltage. It's creative, to say the least. Took me some time to look through it. Do you see it?
Jan Didden
Right after postting that dawned on me!
After looking at it a few times........( first couple made me think someone made a mistake drawing it) The circuit is a unity gain volt regulator with 5 V zener references. Instead of referring the Zeners to ground and the noninverting input, they are referred to the output. The circuit is still a classic unity gain feedback regulator but with the output voltage shifted by the zener voltage so that both inputs of the op amp are at ground instead of 5 volts. This biasing of the Zeners of the other polarity regulator output gives the Zeners a cleaner bias source and the regulator a better PSSR. In a nutshell, they wanted AC unity gain for the regulators with a common mode voltage within the acceptable range for the op amp. It is definitely a tail waging the dog circuit, but the design is intentional. The zeners should have about 5 ma through them for low noise and dynamic impedance. The bias resistor to the other output is around 1K ohms. the circuit could be easily improved with more zener noise filtering.
The 30 minute editing limit gave me the chance to look stupid but I won't ask the moderators to delete the first one. This is definitely different than any way I have seen the refence voltages done but
was definitely done for unity gain feedback and a good common mode
voltage. maybe it will come in handy some day.
After looking at it a few times........( first couple made me think someone made a mistake drawing it) The circuit is a unity gain volt regulator with 5 V zener references. Instead of referring the Zeners to ground and the noninverting input, they are referred to the output. The circuit is still a classic unity gain feedback regulator but with the output voltage shifted by the zener voltage so that both inputs of the op amp are at ground instead of 5 volts. This biasing of the Zeners of the other polarity regulator output gives the Zeners a cleaner bias source and the regulator a better PSSR. In a nutshell, they wanted AC unity gain for the regulators with a common mode voltage within the acceptable range for the op amp. It is definitely a tail waging the dog circuit, but the design is intentional. The zeners should have about 5 ma through them for low noise and dynamic impedance. The bias resistor to the other output is around 1K ohms. the circuit could be easily improved with more zener noise filtering.
The 30 minute editing limit gave me the chance to look stupid but I won't ask the moderators to delete the first one. This is definitely different than any way I have seen the refence voltages done but
was definitely done for unity gain feedback and a good common mode
voltage. maybe it will come in handy some day.
I took your advice (thanks) and replaced the 5.6k resistors for 1k (993ohm) and the output raised to 4.6v. It's almost good enough.
I also changed the 2.4k resistors for 100ohm and this had no apparent effect on the output. Measured the voltage on the inputs of the opamp and got 0.1mV on one side 0.4mV. Looks good.
I wonder how to get the last .5V out of this thing. The zeners are 5.1v 5% so the lowest I should get is 4.85v.
I may try a lower value for the bias resistor maybe about 600ohms?
I also changed the 2.4k resistors for 100ohm and this had no apparent effect on the output. Measured the voltage on the inputs of the opamp and got 0.1mV on one side 0.4mV. Looks good.
I wonder how to get the last .5V out of this thing. The zeners are 5.1v 5% so the lowest I should get is 4.85v.
I may try a lower value for the bias resistor maybe about 600ohms?
I wonder how to get the last .5V out of this thing.
Find out what output voltages of the op amps are. You probably should try a 9 Volt zener (or 8 to 10 volt) it isn't critical. If the saturated drain current value (Id at Vgs = zero) of the K246 current sources across the 2.4 K resistor to the op amp output gives too much voltage drop, the op amp will run out of output swing. For instance the op275 will only swing within a couple of volts from the rail. For an Ids of 6.5 mA (the top of the GR bias range for the 2SK246) the op amp output is sitting around -10 volts. A high impedance like a 2.4K resistor is also degrading the regulator output impedance and a zener will work much better.
Unless you are getting the actual zener voltage at the regulators output it is not in regulation. Measure the zeners at the expected current levels out of the regulator circuit and note the values.
Find out what output voltages of the op amps are. You probably should try a 9 Volt zener (or 8 to 10 volt) it isn't critical. If the saturated drain current value (Id at Vgs = zero) of the K246 current sources across the 2.4 K resistor to the op amp output gives too much voltage drop, the op amp will run out of output swing. For instance the op275 will only swing within a couple of volts from the rail. For an Ids of 6.5 mA (the top of the GR bias range for the 2SK246) the op amp output is sitting around -10 volts. A high impedance like a 2.4K resistor is also degrading the regulator output impedance and a zener will work much better.
Unless you are getting the actual zener voltage at the regulators output it is not in regulation. Measure the zeners at the expected current levels out of the regulator circuit and note the values.
If you take 5.6 volts and put through a smaller current you may get 5.0 volts but this is a typical tolerance thing. You have to have 10, maybe more zeners and match them with the current you plan to have. +- 5% and with different current than for what the tolerance is given may give you +- 10-15 % in your case, maybe even more..
If you have 1 mA through the zener I think 5.6 volts will be a good choice but I'll suspect that more current is better. Use as much current as you can "afford" when it comes to power dissipation and how much load you have got. 5-10 mA is good values I think.
If you have 1 mA through the zener I think 5.6 volts will be a good choice but I'll suspect that more current is better. Use as much current as you can "afford" when it comes to power dissipation and how much load you have got. 5-10 mA is good values I think.
Git a reply from China: (relevant text)
----start---
The polarity of K246 please insert like the white silk screen components outline show on the pcb.
The voltage output only -3.74 or other is too low, change the K246 to K246GR, if the voltage is stil un-normal you can change the fet to zener diode.
The pcb of the DAC38 is fully tested and working good.
Hope it may solve your problem.
----end---
As I mentioned before the K246 is marked on PCB as reversed from schematics - Drain/Gate connected together instead of Source/Gate.
----start---
The polarity of K246 please insert like the white silk screen components outline show on the pcb.
The voltage output only -3.74 or other is too low, change the K246 to K246GR, if the voltage is stil un-normal you can change the fet to zener diode.
The pcb of the DAC38 is fully tested and working good.
Hope it may solve your problem.
----end---
As I mentioned before the K246 is marked on PCB as reversed from schematics - Drain/Gate connected together instead of Source/Gate.
Befor I was rudely interupted How about an hour edit window?
"I wonder how to get the last .5V out of this thing."
Find out what output voltages of the op amps are. If the saturated drain current value (Id at Vgs = zero) of the K246 current sources across the 2.4 K resistor to the op amp output, gives too much voltage drop, the op amp will run out of output swing.
For instance the op275 will only swing within a couple of volts from the rail. For an Ids of 6.5 mA (the top of the GR bias range for the 2SK246) the op amp output is sitting around -10 volts. With a 12 volt supply for the op amp you are sitting in the forbidden zone. That's one of the dangers looking at just a schematic, I will bet that a warning about this was in any article that might have accompanied the schematic. At the least, I bet a Y bias group (1.2 to 3 mA) was specified for the 2SK246. The Y bias group would have
also given the highest output impedance for the fet which is desirable for use as a CCS. A high impedance like a 2.4K resistor between the transistor base and op amp output, is degrading the regulator output impedance and a zener will work much better here for low impedance and control of DC voltage at the output of the op amp. You probably should try a 9 Volt zener (or 8 to 10 volt) it isn't critical.
Unless you are getting the actual zener voltage at the regulators output it is not in regulation. Measure the zeners at the expected current levels out of the regulator circuit and note the values.
I hope this is making it clear that the part values come from somewhere before magically appearing on the schematic. The schematic is an snapshot of where the thing wound up after the design was finished. See why there is a space for Revision Number on a formal schematic? Changes to a design almost always occur after the first schematic. Writing the dreaded Engineering Change Order is usually sufficient punishment for mistakes that are carelessness or a lack of foresight by the designer, unless you are writing to fix someone else's mistake, which is often the case.
"I wonder how to get the last .5V out of this thing."
Find out what output voltages of the op amps are. If the saturated drain current value (Id at Vgs = zero) of the K246 current sources across the 2.4 K resistor to the op amp output, gives too much voltage drop, the op amp will run out of output swing.
For instance the op275 will only swing within a couple of volts from the rail. For an Ids of 6.5 mA (the top of the GR bias range for the 2SK246) the op amp output is sitting around -10 volts. With a 12 volt supply for the op amp you are sitting in the forbidden zone. That's one of the dangers looking at just a schematic, I will bet that a warning about this was in any article that might have accompanied the schematic. At the least, I bet a Y bias group (1.2 to 3 mA) was specified for the 2SK246. The Y bias group would have
also given the highest output impedance for the fet which is desirable for use as a CCS. A high impedance like a 2.4K resistor between the transistor base and op amp output, is degrading the regulator output impedance and a zener will work much better here for low impedance and control of DC voltage at the output of the op amp. You probably should try a 9 Volt zener (or 8 to 10 volt) it isn't critical.
Unless you are getting the actual zener voltage at the regulators output it is not in regulation. Measure the zeners at the expected current levels out of the regulator circuit and note the values.
I hope this is making it clear that the part values come from somewhere before magically appearing on the schematic. The schematic is an snapshot of where the thing wound up after the design was finished. See why there is a space for Revision Number on a formal schematic? Changes to a design almost always occur after the first schematic. Writing the dreaded Engineering Change Order is usually sufficient punishment for mistakes that are carelessness or a lack of foresight by the designer, unless you are writing to fix someone else's mistake, which is often the case.
The easiest way to get exactly (more or less) 5V is to replace the zeners with floating integrated 5V references. Search the sites of Analog Devices, LTI, Texas etc, there's tons of them. They are more accurate and have better (lower) dynamic impedance than zeners which means better regulator performance.
For example, the ADR550 will beat the hell out of any zener regulated supply.
Jan Didden
For example, the ADR550 will beat the hell out of any zener regulated supply.
Jan Didden
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.