Strange problem with simple power supply

[rock soderstrom]

Ruud, What DC-DC IC does your board use?

I am not Ruud, but he mentioned it in this thread:

"EDIT 29/08; Today I looked with 'better eyes', the chips are CXW8509 and MAX1771."

 

[rock soderstrom]

The operation with a standard 9V DC wall wart is very tempting.

This of course raises the question of where to get a PE connection? An effects unit could be set up with protective insulation case, but not a tube microphone.

You would have to do it the way Ruud did.
Is there a qualitative difference for B+ whether the module is supplied with only rectified + RC filter or a regulated DC voltage?

 

[RuudNL]

At first I used the +210 output of the DC-DC converter directly, but this resulted in a very noisy signal from a ('real') U67. After adding a 4.7K/47µF filter, there was no audible noise anymore.
Unloaded voltage is now 213V, loaded voltage (with a U67) is now 208V.
(208 V. is the same voltage the original Neumann power supply delivers.)

 

[rock soderstrom]

After adding a 4.7K/47µF filter, there was no audible noise anymore.

That’s totally ok for me. Look what the postman just dropped at my door!

 

[elskardio]

The output voltage of the DC converter drops, so no excessive dissipation in the resistor.

I have swapped the two DC converters; the one I used for the +210V now only delivers 6.3 V and vice versa.
To my surprise this solved the problem. So it seems one DC converter had a problem.

Still it is strange that the 'faulty' DC converter delivered 210 V without a load, and did the same with a 10 mA load, but dropped the voltage when an R-C load was conneted. Well, problem solved it seems...

View attachment 122813

Hi RuudNL,

how did you connect the chassis ground and the DC converter's ground?

Are they tied together? I know some DC converter only works well when the output ground is floating...

Thanks

 

[bremusound]

Hi, i ordered some of these little boards from Ebay, but they put out 6.7V instead of 6.3 (i used a 33 ohm 2w resistor to simulate an EF86 filament)
I found out that R6 is 62k but for 6.3V it should be about 58k. A paralleled resistor (~680k) will give you the right output voltage.

 

[abbey road d enfer]

Hi, i ordered some of these little boards from Ebay, but they put out 6.7V instead of 6.3 (i used a 33 ohm 2w resistor to simulate an EF86 filament)
I found out that C6 is 62k but for 6.3V it should be about 58k. A parallel resistor with about 820K or 910K will give you the right output voltage.

Boards similar to those in post #24 have a trimmer that allows tuning the output voltage by about +/-10%.
C6 is a descriptor for a capacitor. It shouldn't have much action on the output voltage.

 

[bremusound]

Sorry it‘s R6 not C6…
My boards have a trimmer for HV output only…

 

[abbey road d enfer]

Sorry it‘s R6 not C6…
My boards have a trimmer for HV output only…

Do you mean the same board produces HT and heaters simulatneously (out of a single regulator)? Then adjusting HT would also adjust heaters, which you may find inconvenient...

 

[bremusound]

No, it has two regulator ICs, one for B+ (variable output, about 140-250V), and fixed filament output (it says 6.3 V but is actually 6.7V on my boards)

 

[rock soderstrom]

(it says 6.3 V but is actually 6.7V on my boards)

The same here, my 3 boards show all 6.7V independent from B+

Here with B+ =240V, heater current = 0.6A

An additional RC Filter is an good idea and can fix this easy.

 

[elskardio]

An additional RC Filter is an good idea and can fix this easy.

 

[abbey road d enfer]

An additional RC Filter is an good idea and can fix this easy.

That, or finding the NFB resistor and adding a resistor in parallels to adjust the voltage to your liking,
OR,
Living with it. 6.7V is only 6% above nominal. Tubes are designed to withstand at least 20% overvoltage.

 

[Matador]

Interesting: getting a MAX1771 to work with a vertically-oriented multi-turn pot is actually quite a challenge, as this chip has no compensation on the FB pin, and will easily turn into a 50MHz transmitter through coupling into the feedback pot.

Can you tell by looking at the PCB if the feedback pin (pin 3) is bypassed?

 

[bremusound]

CXW8509 is used for the heater output.

 

[RuudNL]

That, or finding the NFB resistor and adding a resistor in parallels to adjust the voltage to your liking,

I think I have placed a 820 K (or was it 680k ?) in parallel with the SMD feedback resistor.

 

[doorunrun]

I've been following this thread and plan to use the module in a Royer Mod power supply.

I'll use RuudNL's Zener diode approach to get HV down to ~100V and CRCR... type of LPF on the HV section for some additional filtering.
But, in early bench tests I'm seeing some HF noise on o'scope cranked wide open (max gain). I didn't try any bypass cap filtering at that time.

Just wondering if anyone else has noticed and dealt with it.

Thanks!

BTW, I plan on using a separate adjustable linear PS for heater voltage.

 

[rock soderstrom]

Interesting: getting a MAX1771 to work with a vertically-oriented multi-turn pot is actually quite a challenge, as this chip has no compensation on the FB pin, and will easily turn into a 50MHz transmitter through coupling into the feedback pot.

Can you tell by looking at the PCB if the feedback pin (pin 3) is bypassed?

I can't see where pin 3 is connected. Can you explain this topic a little more?

 

[abbey road d enfer]

But, in early bench tests I'm seeing some HF noise on o'scope cranked wide open (max gain). I didn't try any bypass cap filtering at that time.

All smps have residual HF noise, which makes the use of additional filtering almost mandatory in any sort of noise-sensitive application.

BTW, I plan on using a separate adjustable linear PS for heater voltage.

I think heater voltage is where the use of an smps makes the most sense, in terms of efficiency.

 

[Matador]

Can you explain this topic a little more?

My first prototypes with the chip howled at high frequency (I can't recall the exact frequency, but it was above 40 Mhz) whenever I used a Bournes 3296W multi-turn pot for adjustment of the output voltage. Stuffing in plain SMD resistors for a fixed output voltage and the problem went away.

Nick DeSmith's page about this chip goes further: https://nick.desmith.net/Electronics/NixiePSU.html

The FB pin is very sensitive to stray EMI - the voltage feedback network should be kept away from high current paths and the connection to the FB pin should be as short as possible. Should your implementation appear to not regulate or not to be adjustable, use a 'scope to have a look at the FB pin (3) - if you see spikes, then that may well be the reason. Putting a suppression capacitor from pin 3 to ground is not a solution - the layout must be correct.

Unlike his recommendation, I did find that a small snubber on the FB pin (a 47pF in series with 4.7k) provided enough HF deemphasis that it became stable with the vertical pot. Flat single turn pots also didn't show this problem. The snubber I got from reading the TPS40210 datasheet from TI, where they expose a compensation pin exactly for this situation.

In fact, I've switched most of my designs over the the TPS chip, because their web software allows for a lot of custom configuration and provides recommended starting values for a variety of output voltages and currents: it will compute the best switching frequency, inductor, and compensation network based on how much ripple you can tolerate, whether you want a shielded inductor, etc.