Testing High Frequencies in B+ supply boost converter (Nooby learns to test)

Help Support GroupDIY:

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
High frequency lessons learned

Some of us still want to use valves and are hooked on the API module format, so there is the issue of B+ supply.  One good way to approach that is to use a high frequency switching boost converter power supply.  These are small, cheap, and if filtered correctly and if the frequencies are high enough, can be made to work well in the 500/GDIY51X environment.

Recently I built the GIX-51X put together by gemini86 in a fairly faithful reproduction of Jacob Erland's Gyraf G9 DIY preamp.  Rodney adapted the preamp to use a boost converter B+ supply. It is a great project, a nice preamp, and I recommend it.

I was originally fooled into thinking the B+ supply was very noisy, but I was wrong because the first thing I learned upon building this is, if you test it with an oscilloscope, your oscilloscope will lie to you! The frequencies are so high, and the radiation from the switching supply is so great that you have to do this to your oscilloscope probe or the ground strap becomes a winding that picks up signals greater than the ones you are trying to measure.
IMG_1745_zps8bdb2c1c.jpg


By doing this and wrapping the oscilloscope leads around a ferrite toroid (and for safety make sure you touch the wire to ground before the test point or otherwise ground the scope) you will be able to read better what is going on.

Basic switching

For those unfamiliar, the basic idea of a switching supply is to convert DC to DC either up (boost) or down (buck) by turning an inductor on and off at high frequency, and then rectifying the pulsed current (AC) created by the flux field collapse that happens when an inductor turns off.  For a B+ supply the architecture would always have a big discrete MOSFET controlled by the controller to handle the voltage. If this sounds noisy, it is, both from a magnetic field and a ripple current/voltage point of view. 

But what Rodney's project showed me is that with filtering and shielding, these noise sources can be managed to levels that are low (below my measurement ability) and in some ways are superior to the linear supplies that we commonly use.  This is because the linear supplies we use have much larger magnetics, which produce huge hard to filter fields and to really filter these frequencies the filter circuitry is physically large. Also, the noise is in the audio spectrum (50/60Hz).  The switching circuits on the other hand are beyond the audio frequency, they are high frequency and effective shields are thin, and effective filters are physically small. 

One problem with switch controllers (plentiful, cheap and powerful) is that the vendors of the "switching controllers" are not thinking so much about vacuum tubes, as they are cell phones and LED lighting supplies! (These generate the highest part sales count). Most DC to DC boost conversion is to take low voltage battery supplied equipment and step it up to a usable level, this allows small batteries with few cells to run powerful things.  And of course in that environment size, efficiency an power saving considerations are paramount, and noise is not (MP3 and Cell calls are not high fidelity).  And in the LED lighting environment the paramount consideration is efficiency, and the noise filtering is just to keep things within FCC limits (I imagine)!

On the other hand, what we want for a B+ supply is a few milliamps of clean High Voltage and Frequencies that unlikely to have harmonics in the audio range. (Efficiency Ha!  We are running heaters!) Ideally we would like high frequency white noise. So when using these controllers, we need to "distort" their original design intent to some extent because instead of Efficiency, Low Voltage and Current we care about Frequency and High Voltage. But it can be done, it is just that most chips are not designed specifically for that purpose.

If you are interested there are some really good resources for learning about this stuff, which I found as I worked on the GIX51X.  I attached a great presentation transcript by Alan Martin of Nat Semi on Boost Converter Design Tips, which is really clear and sorts things out.  There is a video of this out there somewhere but a PDF slides and transcript of his speech are attached.  Good google search terms are "Boost Converter" with or without "Input Filter", "Output Filter" and "EMI".

Rodney has a nice schematic starting place over on the GIX51X thread using the Max1771 although he is looking at alternative chips for future projects.

(I myself looking at "Finger 16" in the GDIY51X rack, and wondering why not put on a B+ Supply (GDIY51XB+). It would easily fit behind the backplane.  This would make more space available on the module, and allow good shielding of the supply, or even the use of linear supplies for purists. Maybe safety concerns are too great... )

 

Attachments

  • Boost Converter Design Tips.pdf
    383.6 KB · Views: 43

JohnRoberts

Well-known member
Staff member
Moderator
Joined
Nov 30, 2006
Messages
20,217
Location
Hickory, MS
When I see that scope probe with a ground wire hanging off it, I have unpleasant flashbacks to past components I have blown up by inadvertently shorting something out with a similar probe.

I understand why we need to keep the ground lead short and close to the probe tip, but advise caution when probing around a power supply.

JR
 

gemini86

Well-known member
Joined
Sep 22, 2008
Messages
2,476
Location
Eugene, OR USA
Bruce, the only reason I would advise against putting HV+ on pin 16 is that when you're plugging in a 500 module, there's no real good way to make sure you aren't off a little in the card edge connector. There's nothing there to align the bottom edge to, so you may end up putting very high voltages on your +48V rail (which lives on pin 15) Bye bye phantom supply. Other DIY forums have considered adding 300V to pin 'M' (front side of the board, above pin 11.). I don't know if any 500 module in existence uses this pin for anything.

VSI-API-Pinouts.JPG


Thanks again for doing the footwork on all the converter testing. It's nice to have a fresh pair of eyes to help make sure I'm not being stupid.
 

Harpo

Well-known member
Joined
Oct 29, 2005
Messages
2,399
Location
Rahden, Germany
gemini86 said:
Bruce, the only reason I would advise against putting HV+ on pin 16 is that when you're plugging in a 500 module, there's no real good way to make sure you aren't off a little in the card edge connector. There's nothing there to align the bottom edge to, so you may end up putting very high voltages on your +48V rail (which lives on pin 15) Bye bye phantom supply. Other DIY forums have considered adding 300V to pin 'M' (front side of the board, above pin 11.). I don't know if any 500 module in existence uses this pin for anything.
For 300V from MIL Std275b you'd need at least 3.17mm distance between traces for cat Ib below 3048m above sea level. The gap in between cardslot pins is only 1.7mm. For cat Ia (no humidity condensating/dry and non conductive pollution) you could get away with 1.58mm, but that might not be realistic at an open air venue... just saying
 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
Harpo said:
gemini86 said:
Bruce, the only reason I would advise against putting HV+ on pin 16...
For 300V from MIL Std275b you'd need at least 3.17mm distance between traces for cat Ib below 3048m above sea level. The gap in between cardslot pins is only 1.7mm. For cat Ia (no humidity condensating/dry and non conductive pollution) you could get away with 1.58mm, but that might not be realistic at an open air venue... just saying

The geometry of the connector form puts the contacts "around the corner from each" in the connector,  I don't know what this does to the MIL spec, but the Dialectric Withstanding Voltage is 1800V AC at Sea Level.

I but I agree that lethal voltages on an open backplane is not such a great idea.
 

emrr

Well-known member
Joined
Apr 12, 2006
Messages
7,785
Location
NC, USA
You'd want micro-fusing on each slot to keep lethality minimized, if you had B+ for 11 modules.  That's not really what worries me though, it's all about the potential to hit the adjacent slot when hot-plugging, which we aren't supposed to do, but our intern might.  Nothing quite like dumping 250V onto 11 condenser mics via the phantom path.  How do you block that? 
 

bruce0

Well-known member
Joined
Jan 24, 2010
Messages
1,065
Location
Boston
In my book, minimum lethality is as bad as maximum lethality!  The key even with these single tube modules is that the the case is grounded, and no fingers can get in.

I didn't know anybody really hot plugged these things.  The edac contacts are only rated for like 1000 insertions (I didn't find the spec).  Wouldn't it be better to build more racks?

 

gemini86

Well-known member
Joined
Sep 22, 2008
Messages
2,476
Location
Eugene, OR USA
I thinka lot of people hot plug. It's one of the selling points. If you're not using a module constantly, swap it out for something else. Now, i don't think anyone was thinking about high voltage when they designed these racks.
 

Potato Cakes

Well-known member
Joined
Jul 1, 2014
Messages
1,876
Location
Nashville, TN
gemini86 said:
Bruce, the only reason I would advise against putting HV+ on pin 16 is that when you're plugging in a 500 module, there's no real good way to make sure you aren't off a little in the card edge connector. There's nothing there to align the bottom edge to, so you may end up putting very high voltages on your +48V rail (which lives on pin 15) Bye bye phantom supply. Other DIY forums have considered adding 300V to pin 'M' (front side of the board, above pin 11.). I don't know if any 500 module in existence uses this pin for anything.

Not to completely deviate, but I actually have 2 VSI Audio tube preamps w/LA2A style compression, each in a single 500 series format that require 300V. I've been wanting to build a 2 channel chassis for them since I obviously can't use them in a regular rack and the one from them has too many noise issues to use. The diagram you posted just made my day! Thanks!

Carry on...
 
Top