Odd 'ripple' in regulated Filament supply

shop deals on amazon
Advertise with us
Shop deals on ebay
GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
amplexus

amplexus

Well-known member
Joined
Jan 11, 2018
Messages
332
Location
Ontario, Canada
So i've got a HV and B+ supply PCB. It's my own layout, but nothing particularly far from the normal application guides.

On the filament supply I'm using an LM338 to feed the heaters of several preamp tubes at ~6V. Vin = 9VDC, Vout is set for 6VDC. The draw when loaded is 3.5A and the chip is well heatsinked to the chassis, laminated with an aluminum bar. I'm getting a weird half wave periodic 'ripple' in the output that is manifesting itself in the audio path at higher gains. Adding additional filtering to the output rail makes very little difference up to something silly like 10,000µF where it only reduces it by around half, which seems odd.

Is this just a function of pulling that much current from the reg? The I-O differential is low, and its bolted to a like 100g hunk of aluminum. All the parts are new, 105C nichicon PW caps... Circuit and scope below.

(Ignore the parts notes on the right hand side. The Transformer for the LV is a VPT18-2780 for 9.0VAC @ 5.56A. Diodes are 4007s)
IMG_2972.jpeg

Screen Shot 2022-07-06 at 12.59.48 PM.png
 
The problem is most probably the insufficient Vout-Vin difference for LM338. Try to use Shottky diodes in the rectifier and increase the values of capacitors C1 and C3. Or find an LDO IC instead of the LM338. Could you post second trace with Vin voltage on LM338?
 
@JohnRoberts @moamps This seems to be the issue- there's enough unreg'd ripple to pull it below the drop out threshold in a cycle. Increasing the pre reg filtering by clipping in another 10kµF helps. My diode bridge is also running a bit warm, so I'm going to up that to a 10A bridge to keep things cooler and with more headroom for this application.

Helps to talk these things out! Cheers!
 
I would just like to add a couple of observations here, I hope no one will blame me.

I noticed that you mentioned that 1N4007 diodes were used and that the output current from the power supply is 3.5A. 1N4007 is a diode rated for 1A cont., which is significantly below 3.5A. In your case, it was probably saved by the fact that in the power supply that diode is loaded with impulse current, for which that diode is rated about 30A. But as you noticed, the dissipation on diodes is still very high. In any case, in this example, more powerful diodes should be used, and I would go a step further and use schottky diodes, which would reduce the dissipation on the diodes by about 50%. I would use say MBR1035 which can be placed in a 1N4007 footprint with a bit of lead bending.

https://hr.mouser.com/ProductDetail...s/VS-MBR1035-M3?qs=MLItCLRbWsx2ZKJdRI%2Bcxg==
As for the regulator, I would use the LDO regulator LD1084V

https://hr.mouser.com/ProductDetail/STMicroelectronics/LD1084V?qs=sepekKm5O7k2pAs/CJPjzQ==
which would certainly give better characteristics than the LM338 in this application.
 
moamps said:
I would just like to add a couple of observations here, I hope no one will blame me.

I noticed that you mentioned that 1N4007 diodes were used and that the output current from the power supply is 3.5A. 1N4007 is a diode rated for 1A cont., which is significantly below 3.5A. In your case, it was probably saved by the fact that in the power supply that diode is loaded with impulse current, for which that diode is rated about 30A. But as you noticed, the dissipation on diodes is still very high. In any case, in this example, more powerful diodes should be used, and I would go a step further and use schottky diodes, which would reduce the dissipation on the diodes by about 50%. I would use say MBR1035 which can be placed in a 1N4007 footprint with a bit of lead bending.
Click to expand...
I believe you misunderstood me, I should have been clearer- The unmarked diodes on the schematic are 4007's. The rectifier is a 6A FWB, 800V 4 lead monolithic GBL type as noted.
moamps said:
As for the regulator, I would use the LDO regulator LD1084V

https://hr.mouser.com/ProductDetail/STMicroelectronics/LD1084V?qs=sepekKm5O7k2pAs/CJPjzQ==
Click to expand...
I think I had originally used the 338 because I was using an 18V xformer series wired, with much lower current demands- but due to the current requirements of this specific project I parallel wired to get the 5.5A at 9VAC. I suppose flipping out to the 1084 for this one is a reasonable change given how cheap they are.
 
This is almost certainly dropout, try lowering the set output voltage down 0.5V (to 5.5V) or so and see if the ripple goes away. You didn't specify if the 9VDC was measured under full load.

The datasheet says that typical minimum dropout at 3.5A is close to 2.5V. If your 9VDC is measured unloaded then it's possible to sag down to 8.5V and then you might start getting into trouble. I think 3V *worst case* (at maximum current) is safe but you are operating pretty close to these levels with 9VDC in. You can scope the input-to-output difference (measure the voltage across the regulator on your scope) and see if the dropout is ever less than 2.5V. I'm guessing it approaches 2.5V at the points where your output voltage dips down.

Lastly, adding output C doesn't do much because the output impedance of the regulator is so low. 10uF on the ADJ pin is far more effective than 10,000uF on the output pin.
 
As Matador wrote, it is almost certainly drop out: the scope display is characteristic of regulator drop out. The input voltage is falling below the minimum voltage for regulation, as shown by the raw dc through-put on the scope. This occurs if the transformer is not rated for the correct voltage output at the desired current, or when the filter capacitors are undersized. The filter cap values shown seem quite adequate, so one must infer that the transformer doesn't meet the demands of your circuit.
 
Doc Morgan said:
As Matador wrote, it is almost certainly drop out: the scope display is characteristic of regulator drop out. The input voltage is falling below the minimum voltage for regulation, as shown by the raw dc through-put on the scope. This occurs if the transformer is not rated for the correct voltage output at the desired current, or when the filter capacitors are undersized. The filter cap values shown seem quite adequate, so one must infer that the transformer doesn't meet the demands of your circuit.
Click to expand...
I've got a 2x15V @ 100VA toroid here that I'm going to knock in in it's place. I was trying to keep the voltage I-O differential low to keep dissipation under control, but obviously TOO low!
 
I can second moamps's recommendation of the LD1084: if memory serves, the dropout at 5A was on the order of 1.2V. Load regulation was worse (no free lunch), however for a heater supply it (likely) doesn't matter much.
 
amplexus said:
So i've got a HV and B+ supply PCB. It's my own layout, but nothing particularly far from the normal application guides.

On the filament supply I'm using an LM338 to feed the heaters of several preamp tubes at ~6V. Vin = 9VDC, Vout is set for 6VDC. The draw when loaded is 3.5A and the chip is well heatsinked to the chassis, laminated with an aluminum bar. I'm getting a weird half wave periodic 'ripple' in the output that is manifesting itself in the audio path at higher gains. Adding additional filtering to the output rail makes very little difference up to something silly like 10,000µF where it only reduces it by around half, which seems odd.

Is this just a function of pulling that much current from the reg? The I-O differential is low, and its bolted to a like 100g hunk of aluminum. All the parts are new, 105C nichicon PW caps... Circuit and scope below.

(Ignore the parts notes on the right hand side. The Transformer for the LV is a VPT18-2780 for 9.0VAC @ 5.56A. Diodes are 4007s)
View attachment 95961

View attachment 95962
Click to expand...
Try a resistor between the input caps like a pi filter and give the regulator some help
 
Murry said:
Try a resistor between the input caps like a pi filter and give the regulator some help
Click to expand...
This will exacerbate his problem, as the output current will drop the input voltage by I*R and make the dropout problem even worse. :)

There are a handful of solutions (from easy to hard/more expensive):

1) Do nothing and drop the output voltage: heaters are 10% tolerant (at least) so 5.5V will work just fine.
2) Switch to a low dropout regulator (if it's pinout compatible should be easy).
3) Find a transformer that has slightly higher voltage.

That being said: 9VAC through a bridge rectifier should be outputting something near 11VDC, not 9VDC, especially with that amount of bulk capacitance....what does the VIN waveform look like?
 
"I'm getting a weird half wave periodic 'ripple' in the output that is manifesting itself in the audio path at higher gains." (emphasis added.)

As others have said this waveform is symptomatic of classic regulator drop-out.

I also second the notion of superimposing the input waveform of the filtered DC onto the output scope trace to see what's going on.

What I find bothersome is that the ripple is half wave and as Matador has pointed out the raw DC voltage seems low.

Are you sure that you have a full wave bridge? I know that you have installed a FW bridge but are you sure one of the diodes isn't open?

Your regulator drop-out voltage waveform should be 100 or 120 Hz.
 
Matador said:
There are a handful of solutions (from easy to hard/more expensive):

1) Do nothing and drop the output voltage: heaters are 10% tolerant (at least) so 5.5V will work just fine.
2) Switch to a low dropout regulator (if it's pinout compatible should be easy).
3) Find a transformer that has slightly higher voltage.
Click to expand...

There are some other solutions :), such as:

- put chokes in the game
- remove the regulator and use only RC networks
- use a capacitor multiplier instead of a regulator
- throw everything away and use some green SMPS
- come on, use the AC voltage to heat the tubes, they are mostly designed for that. How is your preamp more demanding than, say, the V76?;)
 

Latest posts

Back
Top