Warm Audio WA-67 - Teardown

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.
One of the cost advantages of dual bobbin xfmrs is that there is no need for insulation between primary and secondaries.
Of course there are a few drawbacks like lower efficiency, highe and radiated flux.

Yes, but this was a single bobbin Neve/Marinair output transformer clone. It was basically made as cheap as possible, and just enough to make it work.I might do a teardown thread on that transformer in the future
 
Last edited:
Well, to be fair, what sort of voltage difference would you expect between the two sides? It's not like it's a mains transformer that needs to withstand hundreds or thousands of volts...
 
PSU continued...

I replaced R1 and R2 from 22K to 9K each, and now I have 210V to 211V at B+.
Those values seem to work pretty well.
Anyone sees any problem with this?

I actually really wanted to open the power transformer and rewind it for fun, but changing 2 resistors it's so simple, cheap and takes no time.
So if you don't see any problem in the filtering arrangement by changing these 2 resistors I will leave it like this.

A note for people measuring B+ voltage in this mic or in Neumann U67 microphones,
the B+ voltage takes a long time to settle in, if you want to have a better reading you should keep the mic connected for at least 30 minutes to have a stable value.
I noticed that in the first 5 minutes after power up the voltage goes from 203V to 208V, then in the 10 minutes after that it's around 210V, and in the next 15 minutes it will increase a little bit very slowly, the value seemed to have become stable only after 30 minutes of warm up and it stabilized between 210,8V and 211V.

WA67 PSU.png
 
Last edited:
PSU continued...

I replaced R1 and R2 from 22K to 9K each, and now I have 210V to 211V at B+.
Those values seem to work pretty well.
Anyone sees any problem with this?

I actually really wanted to open the power transformer and rewind it for fun, but changing 2 resistors is so simple, cheap and takes no time.
So if you don't see any problem in the filtering arrangement by changing these 2 resistors I will leave it like this.

A note for people measuring B+ voltage in this mic or in Neumann U67 microphones,
the B+ voltage takes a long time to settle in, if you want to have a better reading you should keep the mic connected for at least 30 minutes to have a stable value.
I noticed that in the first 5 minutes after power up the voltage goes from 203V to 208V, then in the 10 minutes after that it's around 210V, and in the next 15 minutes it will increase a little bit very slowly, the value seemed to have become stable only after 30 minutes of warm up and it stabilized between 210,8V and 211V.


wa67-psu-png.88920
The slow ramp time is just part of the overall characteristic of power supplies like this.

Sombeody correct me if I'm wrong, but the zener string at the end of the PSU actually doesn't do anything until B+ hits 212V. The purpose of that zener string is to "regulate" the output of the supply. If Whoops were to further reduce R1/R2 to increase the voltage across C6, then the zener string would act as a regulator of sorts to clamp the B+ voltage at 212V. The upshot would be a "regulated" PSU that hits it's target voltage quickly rather than 30 min.
 
Anyone sees any problem with this?
Nothing very wrong, but as TheJames hinted, there is no regulation from the zeners. One can argue they are not useless, because they actually protect against overvoltage. As I suggested before, making R1 & R2 6.8k would provide better regulation and as a collateral, settling time would be significantly shorter, only seconds instead of minutes.
Now the long settling time is not a huge problem since the audible effects are negligible.
I wouldn't lose sleep over this.
 
Last edited:
According to ohm law, the images in the thread and a few similar scrapped PCBs I have, R1 and R2 have a value of 20kOhm and not 22kOhm.

The small fluctuation of the B+ voltage is not important for the operation of the EF86 electronic tube, but may have an effect on the capsule polarization voltage. If the B+ voltage is 190V, and the resistor divider to obtain the capsule polarization voltage is identical to the original, the current polarization voltage of the capsule is 55V. If the mains voltage drops to 210V, the polarization voltage of the capsule will drop by an additional 5V. A similar situation occurs if the mains supply voltage is higher than the nominal one. This fluctuation can cause the microphone to sound a little different in another place, or at another time of day.

Changing R1 and R2 from 20kOhm to 6k8 is not good idea IMO, because the zener diodes are not brought that way to their optimal operating range, but to the knee of the characteristics, where the highest noise from diodes can be expected. My recommendation would be to short-circuit resistors R1 and R2 (maybe R4 too), which would lead to an increase in zener diode current in the area of good regulation and low dynamic resistance. At the same time, B+ fluctuations due to the change in mains voltage would be completely suppressed.

The biggest problem with this power supply design is that it is amateurishly copied from a (clone) design where B+=120V (C12 clones) is used.
 
Simulation shows that the output impedance of the supply is dominated by the last capacitor, even when shorting R1, R2 & R3, which results in running the zeners at 3mA instead of <1. The dynamic resistance of teh zeners does not change significantly between 1 and 3 mA. Actually running the zeners at low currents helps filtering their noise.
A better arrangement would place the zeners across one of the caps placed before last, so their noise would be better filtered.
 
I may be wrong, but with a string of 5 6k8 resistors, and an input voltage of 250V and an output voltage of 210V, the current is 40/34 = 1.18mA, which, provided that the microphone consumes about 1mA, leaves too little zener diode current.
In conditions when the voltage of the network is even lower, the situation is even worse.
 
According to Whoops, AC is 187.5V or about 265 peak. Simulation says 0.7mA through the zeners, with 0.9mA draw from the mic. I know simulation is not real life, though.
The problem is that the design is so marginal, any deviation from original data results in conceptual variation, such as zenres working or not.
 
The DC voltage measured on the first capacitor is 253V, which most likely means that the mains voltage is quite distorted.
So I have to redo the sim for fiction meeting reality.
EDIT: actually, the loss of peak rectified voltage cannot be ascribed solely to the DCR/leakage inductance, so it seems the core is saturating. I'm too lazy to run a Chan simulation, so I'll leave it there.
 
Last edited:
sorry to digress

Does the power supply have any safety agencies markings UL etc?
'markings' don't mean what they used to:

it is peculiar when you look on ebay or amazon or your online 'marketplace' of choice how electronic products will be listed --and sometimes the physical packaging--with all manner of suspicious certification-looking marks

parts from a reputable vendor may have legit marks but elsewhere it's sure grey. copy-paste graphics is easy but I've seen knock-off switches and wall-warts with CE certification marks cast into the plastic which is not at all the same as UL, but it would not surprise me to see UL market grey market parts. As I understand it CE can mean little to nothing and is self declared by the manufacturer and UL has been tested and documented at cost to the manufacturer.
To a consumer it is a bunch of logos that hint at legitimacy.
 
Thank you so much guys for you input

The slow ramp time is just part of the overall characteristic of power supplies like this.

Don't get me wrong, I don't mind and don't care about the slow ramp,
I justed added that note as to help other people that are measuring U67 PSU's in the future that to get a real B+ measurement one should wait around 30 minutes until things get stable, it's just something to be aware of.
Personally I don't care, in a recording context most of the final takes and recordings will be done in the hours after the first 30 minutes of warm up time.

As I suggested before, making R1 & R2 6.8k would provide better regulation

I'e tried R1 & R2 with 6.8k as you suggested but B+ was reaching 212V, and my goal was to have it around 210V. So I tried 9K resistors and that provides the 210V.
I'm just trying and testing, nothing is final, if it comes to it and it's needed I can quickly install 2x6.8k resistors.
But do we really need regulation here?
U67 PSU's are not regulated and that doesn't seem to be a problem.
I know the Zeners are there with the intention of providing regulation at 212V, but how important is that?

According to ohm law, the images in the thread and a few similar scrapped PCBs I have, R1 and R2 have a value of 20kOhm and not 22kOhm.

For sure, seeing the color code (Red, BLK, BLK, Red) R1 and R2 are definitely 20K,
I made a typo there, I was really tired in the day I traced that schematic and it was pretty late also.
I will rectify it in the schematic.
But the AC and DC voltage measurements are correct.


If the mains voltage drops to 210V, the polarization voltage of the capsule will drop by an additional 5V. A similar situation occurs if the mains supply voltage is higher than the nominal one. This fluctuation can cause the microphone to sound a little different in another place, or at another time of day.

The mains voltage fluctuation will never be as high as reducing to 210V.
Mains voltage in Portugal is quite stable, the minimum I've seen was 226V and the maximum 236v,
but 85% of the times and places it's from 230V to 233V.
Having worked with many U67 microphones in the past, and actually having fixed their PSU's, which are unregulated, and considering also there's not a lot of mains fluctuations in the places the mic will be used I'm really not worried too much about regulation.
Of course if there's some detrimental impacts of the Zeners like added noise when the B+ voltage is around 210V, I may completely remove the Zeners, change their placement or lower the values of R1 and R2.

A better arrangement would place the zeners across one of the caps placed before last, so their noise would be better filtered.

I could easily do that if needed.

The DC voltage measured on the first capacitor is 253V, which most likely means that the mains voltage is quite distorted.
EDIT: actually, the loss of peak rectified voltage cannot be ascribed solely to the DCR/leakage inductance, so it seems the core is saturating. I'm too lazy to run a Chan simulation, so I'll leave it there.

I also found it strange that the voltage from the transformer secondary was 187.5VAC but after the Bridge Rectifier it was 253VDC when I was expecting it to be 265VDC (187.5 x 1,414).
I thought it was strange.
How can I test or measure if the mains voltage is distorted or the transformer core is saturating?
Should I connect it to the oscilloscope?

Would rewind the transformer for an higher secondary voltage solve that problem?

Thank you so much
 
With 253 at the rectifier , 190 at the mic, or 210 at the mic once modified, and 400V caps, the zeners seem pretty much pointless. Likely mindless application of technology without practical observation. MAYBE the zeners do something for you when the PSU is on when the cold mic is plugged into it, but we don't have a history of known Neumann problems from that either....and we have a 160V coupling cap, so.....
 
I think the point that abbey and I are making is that you can achieve a more stable operating voltage by allowing higher than 212V to hit the zener string. If you hit the zener string with 220V, the zeners will knock that down to 212 and give you some actual regulation of the voltage. It's crude, but practical tech. If you want to hit exactly 210V, then find a combination of zeners that hits 210 and swap in. You'll be right on target, you'll be on target in seconds, and if the line-sags, your 210V won't sag with it. As I think moamps pointed out, if you put the zener string in front of C6, you could absorb some of the zener noise and quiet down the PSU a little more.

As a few other people have pointed out as well, the transformer itself isn't quite up to the task of doing the job. It obviously "works" but it's not really a strong sort of working. To make a better supply without much effort, get a better transformer, get a zener string that equals 210V, move it in front of C6, enjoy.
 
Hi TheJames,
I completely understand that hitting the zener string with more than 212V will make it regulate to 212V,
but as I explained I just don't see why that's needed, U67 PSU's have no regulation and I don't see that being a problem...

As for the transformer, I don't need to get a better one, I can just rewind this one if needed.

But thanks for your input and I will for sure consider the options you suggested.
For now I will just like to know how can I test or measure if the mains voltage is distorted or the transformer core is saturating?

Thanks
 
Hi TheJames,
I completely understand that hitting the zener string with more than 212V will make it regulate to 212V,
but as I explained I just don't see why that's needed, U67 PSU's have no regulation and I don't see that being a problem...

As for the transformer, I don't need to get a better one, I can just rewind this one if needed.

But thanks for your input and I will for sure consider the options you suggested.
For now I will just like to know how can I test or measure if the mains voltage is distorted or the transformer core is saturating?

Thanks
Can't say I've ever had the need to monitor the waveform of an AC outlet, but assuming you have a scope and the scope can handle the voltages, just strap it across the outlet, and then the secondary of the transformer. All safety precautions need to be heeded, and proceed at your own risk.

You're right, you don't need to regulate the microphone voltage...but you seem determined with hitting 210V on the B+ for the microphone. Just suggesting how to do it with what you've got.
 
Last edited:
Back
Top