Help with radiating 60 cycle hum introduced at the input Xfo

[midwayfair]

I just finished a pair of preamps (based on the Hamptone FET preamp but less gain) that has some low-frequency hum the my DAW reads as much as ~-48dB at full gain (~62dB),  when loaded by an e609 dynamic, which I think is unacceptable and it's much more than other preamps I have around. If I move the preamp box away from my computer/interface or sit it onto its side (90 degree angle), the hum drops to -55 to -60dB and is only a little louder than hiss, which isn't perfect but is certainly very good. I tested for hum with the other pres I have (an ART MPAII, the pres in my Scarlett interface, and a Baby Animal Neutral pair I finished recently) and those are all closer to -70dB for the 60 cycle hum.

The input transformers are Edcor XSM, 600:10K. Here are the transformer docs: https://www.edcorusa.com/xsm600-10k

I did the following to verify that it was definitely being generated at the input transformer:

• There is no hum using the DI, e.g. with a guitar (except from the guitar pickups with single coils) -- this bypasses the input transformer.
• When I turn the pad down, the hum gets quieter(I get more hiss, though, as expected). This does not change the signal:noise, however: I confirmed it confirmed by recording a sample with the -18dB pad in and out and then applying +18dB of gain. The hum is exactly the same level.
• If I ground the input connection to the PCB, there is no noise at all even with the gain at maximum, except a very very low level of hiss, so the hum must be created prior to the PCB input connection.
• Reducing the input capacitor filters out some of the hum.
• Turning down the gain decreases the amount of hum (but does not change the signal:noise).
• If I ground the center tap of the Input transformer. This, again, doesn't change the signal:noise ratio.
• The placement of the transformers inside the box itself doesn't matter except as it relates to anything radiating.

Other things:

[list type=decimal]
[*]The hum is less (not just better signal:noise) with any sort of active mic, but it's still there. It's around -60dB at worst with a small diaphram condenser with 150R termination (I'm using my DAW for the analysis). I really have no idea what that indicates.
[*]The box is a hammond aluminum 12x8x2 chassis. I've built guitar amps in these that are quiet, but those had shielded transformers. Everything is well grounded, and there aren't any symptoms of ground loops (i.e. no buzzing).
[*]There ARE harmonics from the hum at 120Hz and 180Hz, but they are minimal. I've already replaced filtering caps, but it's my understanding that if it were bad filter caps the hum would be most prominent at 120Hz, not 60.
[*]The transformers are open, so I wrapped one of the input transformers in foil and then grounded the shield to the chassis, but this does not change the hum at all. I'm not sure I could effectively shield a transformer anyway, because this means it's a magnetic field problem and the aluminum isn't a magnetic shield, right?
[/list]

May or may not be relevant:

• When I first wired it up I messed up and had the center taps grounded on the primary (derp), which put DC across the coil. I did not identify any damage -- no heat or smoke or anything like that -- and I corrected the mistake quickly.

Is there anything I can do about this? I've read more than once that the only thing that can really be done about radiating hum is to move the device away from whatever it's picking up interference from, but that's less than ideal.

I have nothing else with transformers and no other transformers to test with.

Do I have to just get something with a humbucking coil? I'm having a hard time finding those. Cinemag's CMMI-8C or  CM-75101A both have the right ratios but I can't figure out if they're humbucking. None of the Jensen transformers seem to have a primary above 150R, which I don't really think is high enough.

 

[Brian Roth]

Mic transformers are almost always enclosed in a mu-metal "can".

Bri

 

[midwayfair]

Mic transformers are almost always enclosed in a mu-metal "can".

Bri

Bugger, I was afraid of that.

Any suggestions other than the Cinemag and Carnhill Xfos for a 600:10K (or 600:15K)? Or is the Jensen 3k7 okay? (There are some JE models on Ebay for decent prices.) It's only 150R primary, which should be okay for any active mics ... about the only things I have around the house that won't be happy with it is an e609, which wants a 1K [!] impedance.

 

[PRR]

You can *try* a short length of iron pipe with pipe-caps.

That tranny is 1.64" tall. 1-1/2 iron pipe (1.61"ID) will be too tight. You might get it by removing the mounting bracket. 2-inch pipe works easy, but is big: 2.375" outside and more over end-caps.

 

[PRR]

> aluminum isn't a magnetic shield, right?

It is at radio frequency. Magnetic fields induce electric fields which oppose the magnetic field.

In the audio band, and especially at power frequency, this effect is almost invisible.

Cheap soft iron (pipe) is much less effective than layers of fancy Mu metals. However it is much cheaper and available locally.

 

[CJ]

you have a steel core transformer with no shields,

you need a shielded nickel core transformer, humbucking would be a plus,

who makes those?  everybody.  UTC, Peerless, St Ives, Marinair, Triad,

how much money do you want to spend, that will tell you which model to get,

you could pick something up on evilbay for cheap, like a DuKAne or UTC ouncer,

Stancor prices are reasonable and they make some good models,

is this for mic or line input?

we could wind you a custom 1:4 Neve input, humbucking, mu can,  but that' s gonna be 40 clams,

 

[ruffrecords]

All the evidence points to magnetic induction. You need to enclose the transformer in a magnetic short circuit and as ha already been mentioned, the best material for this is mu metal. However, soft steel is a reasonable substitute. As PRR pointed put, the problem is finding a suitable box. Some time ago I had a problem with the magnetic field from a toroid mains transformer and I found  a suitably sized steel box on eBay. There seems to be quite a few of these around  - many of them are used for electrical installation work. This is the one I bought but I am sure you will be able to find other sizes:

http://www.ebay.co.uk/itm/Galvanised-Adaptable-Steel-Box-Electrical-Cable-Enclosure-100x100x50mm-4x4x2-/141003623809?hash=item20d4788d81:g:zdwAAOxy0aBRnhbC

Cheers

Ian

 

[ioplex]

Make sure you don't ground the primary of the transformer. The primary should be "floating". Pin 1 of the XLR input should be connected to chassis ground and that's it. Pins 2 and 3 should be connected to the primary of the transformer and to nothing else. And make sure that the ground connection from the input of the gain stage has a separate connection all the way back to the grounding point at the filter caps.

Grounding is the #1 source of hum in audio gear. It has to be done it just the right way or you will get noise. If the input of a gain stage is grounded to a noisy grounding point, that noise will be amplified by the gain stage.

Google "pro audio star grounding" or similar.

However, your transformer is quite wrong for a microphone so it is very possible that even if your grounding is exactly right, you'll still get hum. The transformer should be the type that comes in a metal can (usually a short capped tube) with pins / wires not just for the metal can itself but for the internal winding shield(s). For example, right now there are 2 Jensen JE-13K6 on Ebay for less than $40 USD shipped. That's a very good transformer and a decent price.

 

[JohnRoberts]

Years ago when I was tooling up some mic transformers (for inexpensive install gear) the Chinese vendor we were working with didn't know what mu metal was.    We eventually got it sorted.

When you put mic preamps transformers inside the same chassis with a 70-100V output transformer you need effective shielding.  8)

JR

 

[midwayfair]

Make sure you don't ground the primary of the transformer. The primary should be "floating". Pin 1 of the XLR input should be connected to chassis ground and that's it. Pins 2 and 3 should be connected to the primary of the transformer and to nothing else. And make sure that the ground connection from the input of the gain stage has a separate connection all the way back to the grounding point at the filter caps.

Grounding is the #1 source of hum in audio gear. It has to be done it just the right way or you will get noise. If the input of a gain stage is grounded to a noisy grounding point, that noise will be amplified by the gain stage.

Google "pro audio star grounding" or similar.

However, your transformer is quite wrong for a microphone so it is very possible that even if your grounding is exactly right, you'll still get hum. The transformer should be the type that comes in a metal can (usually a short capped tube) with pins / wires not just for the metal can itself but for the internal winding shield(s). For example, right now there are 2 Jensen JE-13K6 on Ebay for less than $40 USD shipped. That's a very good transformer and a decent price.

Thanks for the advice.

As I said, I had the center tap on the primary connected by mistake briefly but fixed it quickly when I re-read the pin names on the xfos

I'm confident that the low frequency hum isn't a grounding issue in this case: I can create a whole other set of problems (buzz) by messing up my ground scheme. I actually had to use three separate chassis grounds to fix ground loops in this box: One at the DC input jack filter cap, and the ground pin of both input XLR jacks. A single star ground with minimal wire runs still buzzed a little.

There's actually a bunch of Jensens on E-bay right now from someone about 30 minutes away from me. My only concern with those is that they won't work well with a couple of my microphones. I tested the 150R input impedance on the ART and it was super dark with the e609 and my ribbon wasn't happy with that at all. I have other preamps I can use with both of those, but a 600R input keeps pretty much everything I have happy (even though the e609 sez 1K).

EDIT: Meant to thank Ian for the link to the steel chassis. I had a heck of a time locating something like that. Since I have a couple leftovers I might try to make that work with a single pre.

 

[ioplex]

There's actually a bunch of Jensens on E-bay right now from someone about 30 minutes away from me. My only concern with those is that they won't work well with a couple of my microphones. I tested the 150R input impedance on the ART and it was super dark with the e609 and my ribbon wasn't happy with that at all. I have other preamps I can use with both of those, but a 600R input keeps pretty much everything I have happy (even though the e609 sez 1K).

Note that the impedance listed in product documentation is just the impedance at which they took measurements. So if the primary is advertised as 150 ohms, that corresponds to a specific load on the secondary. For example, if you took a transformer listed as 150:600 and loaded the secondary with 4K instead of 600, the primary would have an input impedance of more like 1K (although at some point, if you drift too far from the load it was designed for, the frequency response will start to become skewed in undesirable ways).

Also, note that a JE-13K6 has a primary impedance of 1.5K with a 35K load on the secondary.

 

[midwayfair]

There's actually a bunch of Jensens on E-bay right now from someone about 30 minutes away from me. My only concern with those is that they won't work well with a couple of my microphones. I tested the 150R input impedance on the ART and it was super dark with the e609 and my ribbon wasn't happy with that at all. I have other preamps I can use with both of those, but a 600R input keeps pretty much everything I have happy (even though the e609 sez 1K).

Note that the impedance listed in product documentation is just the impedance at which they took measurements. So if the primary is advertised as 150 ohms, that corresponds to a specific load on the secondary. For example, if you took a transformer listed as 150:600 and loaded the secondary with 4K instead of 600, the primary would have an input impedance of more like 1K (although at some point, if you drift too far from the load it was designed for, the frequency response will start to become skewed in undesirable ways).

Also, note that a JE-13K6 has a primary impedance of 1.5K with a 35K load on the secondary.

Well, that makes my life easier. (The inputs in this box are currently ~80K at the lowest for the BJT input and ~280K for the FET input, in part because I wasn't totally sure whether I'd be using a 1:8 or 1:10.) I thought there was some sort of intrinsic rating, like with guitar output transformers specifically spitting out a handful of Ohms and getting destroyed with the wrong load. A lot of datasheets provide the actual resistance across each secondary or primary, for instance.

I really don't have a good handle on transformers, so I probably should have done some more reading before I committed this project to a box. How do I know how far away from "the load it was designed for" can I get anyway?

I'm waiting to hear back from CJ, but I might pick up the Jensens regardless.

 

[emrr]

Reread the Jensen data sheets.  150 is the intended SOURCE Z, not the transformer Z. 

 

[ruffrecords]

The thing to remember about transformers is that they do what it says on the tin - they transform. If you have a 1:10 transformer and connect a 150 ohm resistor across the primary, the transformer transforms it so from the secondary side it looks 100 times bigger i.e. 15K. Similarly, if you connect a 150K resistor across the secondary. the transformer transforms it so it looks like 100 times less from the primary side i.e. 1500 ohms. Any 1:10 transformer will do this and it will do it with any values of resistor.

What this means is there is no such thing as a specific impedance for a transformer. However, practicalities dictate that a particular transformer will only transform over the entire audio bandwidth for a specific range of impedances. The primary inductance of the transformer generally dictates the highest impedance that can successfully be transformed at low frequencies. Stray capacitance limits the performance of the transformer at high frequencies. Between these two limits the transformer manufacturers steer a course.

For a mic pre, it is common to make the impedance looking into the primary about 1500 ohms. As in the example above, for a 1:10 transformer, this can be achieved by a 150K resistor connected across the secondary. Typical 1:10  Jensen, Cinemag, Sowter and Lundahl transformers achieve this. The impedance of the connected mic can be anything up to about 600 ohms and be successfully transformed. A condenser mic with a 50 ohm output impedance, or a 150 ohm dynamic or ribbon will all 'see' the 1500 ohm input impedance.

Cheers

Ian

 

[ioplex]

How do I know how far away from "the load it was designed for" can I get anyway?

You would probably have to use an audio analyzer to measure the frequency response and distortion. Unfortunately a good audio analyzer is super expensive. But a PC audio input with the right software is usually good enough for comparative measurements. I used to use DSSF3 from ymec.com for this.

In practice you should just get the right transformer and use the prescribed load. Any mic pre worth it's weight in donuts will use a good shielded input transformer designed for a microphone.

 

[CJ]

those old UTC's had 3 mu cans seperated by 2 sheets of copper, and the whole enchillada was dropped into a steel can,

that was when you would have the mic pre sitting next to the tank circuit i guess,

 

[CJ]

spinning some coils, Neve type, Litton and Hurst lam style,  LH 2023 DU

so we are gonna take a Neve mic input with 499T #36 : 1045T # 39 per coil and modify the turns ratio into 1:4,

gonna need room for more sec wire, so we cut the pri down to 300T #36 and spin 1200T of #38 since we want DCR to stay the same, had just enough room, used #38 for wire shields, saves messin with copper sheets which involves taping them up carefully so they don't short out, and then running leads of fine wire so you don't have a big lump in the coil,

hopefully the shields will improve  noise rejection and maybe sq wave ringing,

pri DCR will be lower which is better as that mic signal is small and DCR loves to eat small signals for breakfast, pri inductance gonna drop? no way, not with this modern hi octane stuff they are stamping in Korea,

primaries were both forward wound so we have a dual coil hookup of S-F-F-S which spells Humbucker in French,

running one xfmr with a reverse wind on one coil, the other xfmr with both fed wind, gonna see the difference if any in inductance, freq response and sq wave ring, exciting isn't it? better than watching the season finale of Friends, 

 

[CJ]

all done winding, over to the lam stuffing table, 80 Supermalloy,

winds are twisted onto terminals, will do hookup after we stuff so the lams act as a paper weight for soldering, no taps brought out, so we connect F to F on one coil and  the secondaries on the rev wind xfmr will go F-S-F-S, hooking leads to outside F and S,

 

[CJ]

laminated, jumpered, and shield wires soldered to the core, now we can run a bigger wire off the other end of the core so that we don't stress those small shield wires, bag o leads and we are ready to roll,

next up, the meat and potatoes of the show, test time! 

 

[CJ]

leads done, found an old st ives foil wrapper, soldered to core, have to look for another one or go down to the stained glass shop,

DCR - Pri both coils in series: 31 ohms - stock Marinair T-1454    47 Ohms
DCR - Sec both coils in series: 247 ohms- stock Marinair T-1454    263 Ohms

Pri L = 23.5 Henries at 20 Hz at 500 mv, this offers up a 3K load to a 150 ohm mic at 20 Hz, plenty good, i mean who sings at 20 Hz, trace Adkins? "Left..., left..., left right left..."  ;D