Seperate USB Data and Power

[user 133392]

As Thor mentioned ,fundamentally , by design most SMPS are unsuitable for audio as they leave the factory , something to do with the approval agencies /safety etc .

I cannot measure a difference with my boards using a linear supply or the TDK CUT35-5FF switcher. https://www.mouser.com/c/?q=CUT35-5FF
There is also a 75W CUT75.

The TDK are one of the best bipolar output switchers I've found that are commercially available.
It may not be the ultimate but I can buy it and do stock them for sale during the times Mouser, Digi-Key et al are out-of-stock.
Unfortunately they are not available in enclosed desktop brick form factor only open frame or enclosed open frame.

I can't say the same however for the Meanwell GP25/GP50 bipolar brick supplies which are noisy.
When they went with "Green Power" mandates the Meanwell brick series turned to crap.

 

[Tubetec]

To be fair to the flash caps , under normal conditions in a disposable camera they might get ramped up to 330v and discharged violently a handfull of times thats it , theres loads of life left in these things ,

I tested a range of different flash caps of varying quality , some cheap Chinese caps , on average measured ten times the ESR that the Fuji camera Japanese made did , all kinds of other parameters on the LCR were showing suspect readings with the cheaper components too .

In useage at 5v there was some of the usual disturbances as the the capacitors formed up, still took some days opperation before all had settled down.


As far as tolerance I made a short note of a sample of the 100uf flash caps I tested ,
its listed here in the pages some where ,
Its was fairly remarkable , but everything was better than +/- 5% , many much closer to the nominal 100uf value , ESR also didnt vary much from sample to sample .
A 10x block of the 100uf's got me around 24 mOhms and 960uf ,

I have also used them at high voltages , upto their specifed maximum and for extended periods of around a month , again no signs of any trouble whatsover ,

 

[Yash]

Overheard in another forum:

the chip crises seems to fade out.

Is that true?

If so, it might be a good time to do get ready for some parts ordering & fun times ahead.

 

[thor.zmt]

I cannot measure a difference with my boards using a linear supply or the TDK CUT35-5FF switcher. https://www.mouser.com/c/?q=CUT35-5FF
There is also a 75W CUT75.

Question is, what are you measuring?

Do you include common noise re Earth in this?

Noise above 20kHz?

The TDK are one of the best bipolar output switchers I've found that are commercially available.

These are basically Lambda, they should rather decent. Looks like single chip per channel, DIP8 is usually quite power limited.

Seems here we get 15W per channel and a bit extra for the dual rail. Some LC filtering - which is good. I'd probably still follow with pretty heavy duty LC filtering on the Audio PCB...

These look to me to require earth on chassis, which can be a problem.

I can't say the same however for the Meanwell GP25/GP50 bipolar brick supplies which are noisy.
When they went with "Green Power" mandates the Meanwell brick series turned to crap.

Yes, that's a problem. Green mode means first lowering frequency and then cycle skipping.

As I remarked a few times by now, USB-C power bricks can be found as 20V/5A max with 48V/5A coming. Simply assume these are noisy AF, but that can be cleaned up.

It is quite easy to turn 20V into all sorts of rails, no need to use modern switchers, the good old MC33063A/34063A will do. Use sensible layout and LC filters to kill noise.

Implement a suitable circuit to drain mains leakage noise currents to earth before they reach the audio circuit.

It all becomes vendor/stock independent.

Thor

 

[thor.zmt]

A 10x block of the 100uf's got me around 24 mOhms and 960uf ,

Not that bad for free. I pay ~1 USD for a 8,200uF/35V Elna Low Z Cap, 18X40mm with low ESR (< 10mOhm).

I have also used them at high voltages , upto their specifed maximum and for extended periods of around a month , again no signs of any trouble whatsover ,

They usually have 500 Hours or so rating, naturally that is 85 degrees and rated voltage and ripple current. Derate any of these and lifespan increases.

If the equipment doesn't run all that hot, they can likely be used in tube equipment with acceptable life span. But again, something to be aware of.

Incidentally, the electronics drivin the Flash usually include a perfectly decent switcher IC and step-up transformer that can make very high voltages at modest current from low voltages. So it may be worth salvaging the whole PCB and reuse with small mods.

Thor

 

[Tubetec]

I'll think I'll stick with the 40VA lump idea KA ,
what I need is a very quiet nicad charger ,


I have been tinkering around with the Flash pcbs , it charges upto 330v from a single 1.5v cell ,
I was able to sucessfully provide HT voltage for a 12ax7 to run , so a few mA's
Typically the charge circuit consists of a few transistors wraped around a switching transformer to charge the cap , then an EHT choke where the cap get discharged through a few turns and ramped upto to KV.
Thats a similar circuit to whats found in the modern units , also turns out the Fuji is very much quieter that the others .

Its proved a handy simple source of high votlage/small current on the bench , without wires and mains and all it brings ,
I'd been meaning for some time to cobble together these components into something more coherent, boxed , with proper filtering of the switcher ,

As I mentioned I have an unlimited supply of used cameras I can get from the film processing store , bags of the Pcbs, capacitors and spare AA batteries .

 

[john12ax7]

The TDK are one of the best bipolar output switchers I've found that are commercially available.
It may not be the ultimate but I can buy it and do stock them for sale during the times Mouser, Digi-Key et al are out-of-stock.
Unfortunately they are not available in enclosed desktop brick form factor only open frame or enclosed open frame.

TDK discontinued a lot of the enclosed supplies they had, quite frustrating.

 

[Tubetec]

In the same experiment with the 12ax7 using the flash board as HT I used a tiny nokia 5v phone charger supply (switched) for heaters , initially it was too noisy , but with the adition of LC filtering all was well , if fact noise from the supplies was entirely negligable with the usual 100k anode load 12ax7 triode circuit .

The beauty for me of the bricks and mortar approach , ie transformer is simplicity , long term reliabillity and spares to maintain things are reduced to a handfull of standard components .
The switchers change and become replaced so your on shifting sand as far as availabillity and performance goes .

 

[Tubetec]

On the subject of powerbanks the Xiaomi PLM18ZM
comes highly reccomended for audio ,
https://www.audiosciencereview.com/forum/index.php?threads/e1da-cosmos-adc.27038/page-56
https://www.audiosciencereview.com/forum/index.php?threads/e1da-cosmos-adc.27038/page-50Its seems people screeming ISO in Ivans ear payed off in the end ,
The new Cosmos ADC iso incorporates a high speed extender and deals with the USB grounding issue in the box .

 

[user 133392]

Noise above 20kHz?

Yes. I've FFT'd the outputs of a number of my boards at 48 kHz FFT BW (SR 96) including both MM and MC preamps and see absolutely no difference at the outputs if the supply is a bench linear or the TDK.

The measurement A/D, which uses an iDefender, is running on an enclosed frame Meanwell RT50-C which attests to it being viable as well.

These are FFTs of my Flat Moving Coil Preamp with a 10Ω source impedance and 61.6 dB gain. (The AudioTester noise measurement analysis bandwidth is 20 kHz.)

The hum in all of these is ambient in the room and physically nulled. The linear, TDK CUT35 and original Meanwell P25 measure almost identically. None have any external filters.

The final FFT is the Meanwell GP25A. The caption shows that the GP25A 5V was not preloaded but I've found no amount of preload on any rail helps it much. It also has a bad line-synchronous hiccup mode which makes filtering impractical and the HF spurs vary in BW depending on load. It appears to modulate its switching frequency to reduce emissions. The GP25A is actually made by EljinTek. (sp?)

I have some broadband measurements here and here.

 

[thor.zmt]

Typically the charge circuit consists of a few transistors wraped around a switching transformer to charge the cap , then an EHT choke where the cap get discharged through a few turns and ramped upto to KV.

I have used IC's with commercially available step up transformers (all SMT) a bit more complex. These had multi-ampere protected switching transistors, so they can output a fair bit or current.

Thor

 

[thor.zmt]

The beauty for me of the bricks and mortar approach , ie transformer is simplicity , long term reliabillity and spares to maintain things are reduced to a handfull of standard components .

Transformers are becoming more rare. As they need a fairly heavy infrastructure, there will soon come a day when the last factory making common transformer lam's goes out of business.

The switchers change and become replaced so your on shifting sand as far as availabillity and performance goes .

The MC34063A I keep banging on about goes back to 1983... That 40 Year's and going strong, not exactly "shifting sand"... And because it is so old, like the old 723 linear regulator it gives a lot of design flexibility for DIY, which can give performance past what more modern items offer.

Thor

 

[thor.zmt]

Yes. I've FFT'd the outputs of a number of my boards at 48 kHz FFT BW (SR 96) including both MM and MC preamps and see absolutely no difference at the outputs if the supply is a bench linear or the TDK.

Reason I asked is that I found a lot of differences with AP2, due to the limited CMRR and various ground loops.

It made me first solve these issues and then create some differential probes to be able to measure these issues using a 300MHz Oscilloscope and 5GHz RF analyser.

Thor

 

[user 133392]

The MC34063A I keep banging on about goes back to 1983... That 40 Year's and going strong, not exactly "shifting sand"... And because it is so old, like the old 723 linear regulator it gives a lot of design flexibility for DIY, which can give performance past what more modern items offer.

Do you find the MC34063A's low and variable switching frequency to be difficult to filter in audio applications? I've used them for Nixie supplies and have suggested that a colleague try it for a single 12AX7 HV supply. I haven't heard back how well the tube supply worked. For my purposes the MC34063 was simple and easy to use.

The µA723 Voltage Regulator Turns 55 This Year - Pro Audio Design Forum

Happy Birthday...

 

[thor.zmt]

Do you find the MC34063A's low and variable switching frequency to be difficult to filter in audio applications?

No, not at all. I used them in first generation ifi products to make dual rails from a single 9V supply, including a MC capable Phono.

Thor

 

[user 133392]

Reason I asked is that I found a lot of differences with AP2, due to the limited CMRR and various ground loops.

When the iDefender and its load are running on externally-injected power and the host's USB ground line is opened by the MOSFET, what is the added shunt resistance between host and device?

 

[JohnRoberts]

I remember using the ua723 in a design about 45 years ago.

JR

 

[thor.zmt]

When the iDefender and its load are running on externally-injected power and the host's USB ground line is opened by the MOSFET, what is the added shunt resistance between host and device?

Pass, I'll look if I have a schematic. Probably 100R.

You can use a multimeter to measure.

Thor

 

[user 133392]

However much it is I've found the iDefender's added impedance to the host ground to be enough to break whatever loop gets setup.

The earlier FFT switcher measurements were balanced with the converter input ground-referred to the DUT. I wouldn't necessarily have that flexibility without the iDefender's added resistance to host ground.

I know you think they may have over-priced it - and it's about a third the price of the 2i2 - but IMHO it provides a lot of value.

 

[thor.zmt]

However much it is I've found the iDefender's added impedance to the host ground to be enough to break whatever loop gets setup.

That was the design goal. I'm happy to know it works for your purpose, it's all I ever wanted.

I know you think they may have over-priced it - and it's about a third the price of the 2i2 - but IMHO it provides a lot of value.

Well, as long as you feel you got VFM who am I to disagree,?

Thor