I have recently had to rebuild a PSU and wanted to try a more modern regulator than the LM317 range.
I bought some LT3080 for low current use and LT 3083 for up to 3A. The 3080 was used with a mini pcb attached to the pins which held the set resistor and the 3 caps, the regulator then fixed to a heatsink. These worked fine but trying to use two 3080's with one as a floating preregulator resulted in a beautifull oscillator.
Attempts to use the LM3083 in the same mode have failed due to my inabilty to stop the 3083 going into oscillation when a load is attached to it.
Im using the prototype at 5v output, and all appears normal until a load is attached.( sinking 200ma.) The 5v output also contains slightly clipped 50mv pk -pk sinewave at around 20Mhz. The spectrum shows unwanted peaks at 1.5Mhz and subsequent 1.5 steps until around 25Mhz.
Set volts reads 5.44 volts and Vout is 5.51volts
Set resistor 100k nominal, bypassed with 0.1 micro farad.
( I have tried higher values to 10 uf)
I have tried various builds with close and wide spacing between components and different capacitors to no avail. It keeps oscillating.
I note the application note shows a ground screen around the set input pin.
Is it possible to use these regulators sucessfully as a DIY build without a proper PCB ?
I have just spotted the LT3080 PCB boards on Group buy/ebay and ordered two in hope of a solution (LT3080 and 3083 have the same 5 pin TO220 case and pin out)
Ianmac
I bought some LT3080 for low current use and LT 3083 for up to 3A. The 3080 was used with a mini pcb attached to the pins which held the set resistor and the 3 caps, the regulator then fixed to a heatsink. These worked fine but trying to use two 3080's with one as a floating preregulator resulted in a beautifull oscillator.
Attempts to use the LM3083 in the same mode have failed due to my inabilty to stop the 3083 going into oscillation when a load is attached to it.
Im using the prototype at 5v output, and all appears normal until a load is attached.( sinking 200ma.) The 5v output also contains slightly clipped 50mv pk -pk sinewave at around 20Mhz. The spectrum shows unwanted peaks at 1.5Mhz and subsequent 1.5 steps until around 25Mhz.
Set volts reads 5.44 volts and Vout is 5.51volts
Set resistor 100k nominal, bypassed with 0.1 micro farad.
( I have tried higher values to 10 uf)
I have tried various builds with close and wide spacing between components and different capacitors to no avail. It keeps oscillating.
I note the application note shows a ground screen around the set input pin.
Is it possible to use these regulators sucessfully as a DIY build without a proper PCB ?
I have just spotted the LT3080 PCB boards on Group buy/ebay and ordered two in hope of a solution (LT3080 and 3083 have the same 5 pin TO220 case and pin out)
Ianmac
Input is 100 uf but close to the PSU resevoir cap of 10000uf.
Output is 100uf.
Caps are from Rubicon and Panasonic.
Have tried different lower values to 22uf and by passing the electrolytics with small value ceramics.
Currently my circuit is as the Data sheet suggests.
Output is 100uf.
Caps are from Rubicon and Panasonic.
Have tried different lower values to 22uf and by passing the electrolytics with small value ceramics.
Currently my circuit is as the Data sheet suggests.
Hi,
it´d be helpful to have a schematic or LTSpice-file and possibly PCB-Pic.
IIuc the single regulator ciruits work fine with both the 3080 and the 3083 and its just your experiment with a Prereg that fails?
Where is the ´Preregs´ Set-Resistor´s gnd pin connected to?
Does it feed into a low impedance node?
Keep in mind that the input of the 308x are high impedance nodes.
Where´s the Set-Resistor current going to flow? (see p11 of 3083 DS)
Do You have enough supply voltage?
The internal ccs needs a bit of voltage to function properly.
This means that if the Vcontrol pin and the Vin are connected together the dropout voltage is 1.25Vmin .... only if Vin and Vcontrol are supplied with different voltages can the low-dropout capabilities be used to full extent.
On p12 of the 3083DS they talk about possible oscillation issues whith long input wires and too low ESR of the caps.
Have You checked the Caps DS and is there any more information than Rubycon and Panasonic and cap-value? 🙄
In general the given information is very insufficient.
I doubt anyone could imagine exactly what You did and how You did things and can´t offer You a decent tip.
Which circuit from which DS for example? LT3083`s DS shows 19 schematics, none of which features a Prereg circuit.
jauu
Calvin
it´d be helpful to have a schematic or LTSpice-file and possibly PCB-Pic.
IIuc the single regulator ciruits work fine with both the 3080 and the 3083 and its just your experiment with a Prereg that fails?
Where is the ´Preregs´ Set-Resistor´s gnd pin connected to?
Does it feed into a low impedance node?
Keep in mind that the input of the 308x are high impedance nodes.
Where´s the Set-Resistor current going to flow? (see p11 of 3083 DS)
Do You have enough supply voltage?
The internal ccs needs a bit of voltage to function properly.
This means that if the Vcontrol pin and the Vin are connected together the dropout voltage is 1.25Vmin .... only if Vin and Vcontrol are supplied with different voltages can the low-dropout capabilities be used to full extent.
On p12 of the 3083DS they talk about possible oscillation issues whith long input wires and too low ESR of the caps.
Have You checked the Caps DS and is there any more information than Rubycon and Panasonic and cap-value? 🙄
In general the given information is very insufficient.
I doubt anyone could imagine exactly what You did and how You did things and can´t offer You a decent tip.
Which circuit from which DS for example? LT3083`s DS shows 19 schematics, none of which features a Prereg circuit.
jauu
Calvin
3080 single regulators I have built work OK.
The pre reg trial had a floating 3080 driving a second 3080.
The pre reg had a set resistor to provide a drop of 4 volts across the 3080. The set resistor was connected to the output of the second 3080 which was set to provide the wanted 9v out. The first pre reg floated on the wanted output of 9v. Its output of circa 13v provided the input to the second reg.
This appeared to work from a DC measurement perspective but did oscillate.
This is along the same lines as I have often used with LM338.
I did not try an LM338 as a pre reg but will.
The 3083 as a single regulator works fine from a DC measurement perspective and only when a load is connected I get 50mv /20Mhz on the output.
I dont have LT spice
I do have enough supply voltage, minimum 4v across the regulator for my trial.
Circuit as fig 1 on P10 on the 3083 data sheet with control and in connected.
Thanks for pointing out the ref to too low ESR caps and long input leads.
I did reduce these to about 2" long but I will now go and try some cheap old electrolytics with a high ESR.
If I dont make any progress today I'll rebuild my original circuit photo it and a circuit.
I had read in another thread that X7R X5R ceramic caps were the ones to use ?
The pre reg trial had a floating 3080 driving a second 3080.
The pre reg had a set resistor to provide a drop of 4 volts across the 3080. The set resistor was connected to the output of the second 3080 which was set to provide the wanted 9v out. The first pre reg floated on the wanted output of 9v. Its output of circa 13v provided the input to the second reg.
This appeared to work from a DC measurement perspective but did oscillate.
This is along the same lines as I have often used with LM338.
I did not try an LM338 as a pre reg but will.
The 3083 as a single regulator works fine from a DC measurement perspective and only when a load is connected I get 50mv /20Mhz on the output.
I dont have LT spice
I do have enough supply voltage, minimum 4v across the regulator for my trial.
Circuit as fig 1 on P10 on the 3083 data sheet with control and in connected.
Thanks for pointing out the ref to too low ESR caps and long input leads.
I did reduce these to about 2" long but I will now go and try some cheap old electrolytics with a high ESR.
If I dont make any progress today I'll rebuild my original circuit photo it and a circuit.
I had read in another thread that X7R X5R ceramic caps were the ones to use ?
Have managed to kill the oscillation on the single 8083.
The output capacitor was changed from an electrolytic 100uf to a tantulum 47uf. It is now stable.
Repeating this on a second 8083 assembly constructed on a wider pitch vero type board with the 8083 wired to the board with short wires ( to allow me to screw it to a case side as heatsink) still oscillated as one increased the load current.
No combination of capacitors I tried effected a cure. I tried removing the short wires from PCB to 8083 and soldered it directly to the PCB. Still oscillates as one increases the load current
I have still to investigate a pre reg set up using another 8083 or LM388 .
I have ordered some PCBs and will wait until I have them and can achieve repeatability in an 8083 build before wasting any more time.
Perhaps back to my favourite Pre reg + reg LM388 pair with the vset resistor replaced with LEDs. ( with more watts to dissipate!)
The output capacitor was changed from an electrolytic 100uf to a tantulum 47uf. It is now stable.
Repeating this on a second 8083 assembly constructed on a wider pitch vero type board with the 8083 wired to the board with short wires ( to allow me to screw it to a case side as heatsink) still oscillated as one increased the load current.
No combination of capacitors I tried effected a cure. I tried removing the short wires from PCB to 8083 and soldered it directly to the PCB. Still oscillates as one increases the load current
I have still to investigate a pre reg set up using another 8083 or LM388 .
I have ordered some PCBs and will wait until I have them and can achieve repeatability in an 8083 build before wasting any more time.
Perhaps back to my favourite Pre reg + reg LM388 pair with the vset resistor replaced with LEDs. ( with more watts to dissipate!)
Not sure, but is the LM388 a regulator, and is your 8083 ref to the LT3083?
If you can provide a picture of your wiring, that would help a lot. Whenever I prototype, I use either copper clad (cut isolation with Xacto) or air wire with very short leads (no more than a few mm).
The 3080 and 3083 may be sensitive to esr and ground returns, as someone else pointed out. Big electrolytics are maybe not the best way to go. Everything I work on these days involves surface mount components mounted to a pcb with gnd planes. I also believe you need a min load current to satisfy the guaranteed specs. It's a subtle thing but these regs are essentially D.C. amplifiers that hv more phase margin with some emitter current out of the npn pass element.
Another thing I'd like to add, the 3080 forces the user to keep leakage very low since the SET current is 10uA. I've used it in the past with an NFET tied to the SET pin to provide shutdown capability, all to find that d-s leakage was introducing an error in Vout. Also solder flux on the set resistor will do the same.
Hope this helps.
Last edited:
Apologies LM 388 is a typo it should be LM338
Ive managed repeatability with the LT3080 and LT3083 using 0.1 spaced Vero strip board, short as possible wiring and trimming the copper tracks so no excess stray capacitance/inductance.
The low set current on LT 3080 is a potential issue I was wary of, as was the minimum load current. The application notes suggest any C used to bypass the set resistor to reduce noise is max 0.1micro f.
My results so far have not quite met my expectations for hum and noise performance and to get my PSU up and running I used my well tried pair of LM338 regulators with one as a floating pre reg.
I did get very promising results using both the LT3080 and LT 3083 fed by a choke input filter but this isnt practical with the 2.5A max current I need, due to Choke cost and physical size.
I am currently looking at a low current supply (50ma) using a choke input filter and a LT3080. This supply will drive the LT3083 Control Pin, with the LT3083 Vin taking the power feed for the actual supply output.
Hopefully this will give me a lower ripple and noise performance.
The Pcb boards I bought are fine and Ok for basic regulators.
Looking through the range of LT regulators I see there are options for paralleling devices with lower noise performance for higher current outputs but I want to play with
the choke input filter idea first.
Thanks for your help.
Ive managed repeatability with the LT3080 and LT3083 using 0.1 spaced Vero strip board, short as possible wiring and trimming the copper tracks so no excess stray capacitance/inductance.
The low set current on LT 3080 is a potential issue I was wary of, as was the minimum load current. The application notes suggest any C used to bypass the set resistor to reduce noise is max 0.1micro f.
My results so far have not quite met my expectations for hum and noise performance and to get my PSU up and running I used my well tried pair of LM338 regulators with one as a floating pre reg.
I did get very promising results using both the LT3080 and LT 3083 fed by a choke input filter but this isnt practical with the 2.5A max current I need, due to Choke cost and physical size.
I am currently looking at a low current supply (50ma) using a choke input filter and a LT3080. This supply will drive the LT3083 Control Pin, with the LT3083 Vin taking the power feed for the actual supply output.
Hopefully this will give me a lower ripple and noise performance.
The Pcb boards I bought are fine and Ok for basic regulators.
Looking through the range of LT regulators I see there are options for paralleling devices with lower noise performance for higher current outputs but I want to play with
the choke input filter idea first.
Thanks for your help.
Last edited:
I was searching around for a solution because I was facing the same oscillation issue with the LT3083 when I came across this post. I managed to get it to work without any sort of oscillation with any load by using ceramic output capacitors (2x22uF).
Some possibly useful datapoints:
- Tantalum output caps (2x47uF): oscillation from about 1.8A load and higher
- Tantalums + Low ESR EL (2200uF): oscillation from about 2A load
- Low ESR EL (2200uF): oscillation from 0.5A load
- Tantalums + 100nF film cap: oscillation from about 2A load
With the ceramic caps the oscillation couldn't be provoked in any condition I tried (up to 3A).
My interpretation is that the datasheet information about 0.5ohm or less ESR for the output capacitor is the absolute highest limit and should ideally be lower than that. So cap choice is quite critical.
Some possibly useful datapoints:
- Tantalum output caps (2x47uF): oscillation from about 1.8A load and higher
- Tantalums + Low ESR EL (2200uF): oscillation from about 2A load
- Low ESR EL (2200uF): oscillation from 0.5A load
- Tantalums + 100nF film cap: oscillation from about 2A load
With the ceramic caps the oscillation couldn't be provoked in any condition I tried (up to 3A).
My interpretation is that the datasheet information about 0.5ohm or less ESR for the output capacitor is the absolute highest limit and should ideally be lower than that. So cap choice is quite critical.
Yep, both input and output caps are within 8-9mm of the respective pins/pads. On the output side they're just after the ballast resistors (it's a dual parallel regulator design).
Unless I am misunderstanding, adding the caps after the ballast resistor will increase the effective cap ESR "seen" by the regulator. May not be the best idea given the datasheet recommendation.
Hi Chris, I'm actually following the data sheet recommendation: check Figure 5, page 13. Ballast resistor value is 10mOhm so there's some margin.
- Status
- Not open for further replies.