DIY signal line transformer ?

[lumanauw]

I've got a problem. In the system, the power amp is placed in the speakers. So, there is a long preamp cable from the master control to each power amp. Many distortions. Hums, strange guitar sound, etc.

At first I tried to solve it by putting balanced xlr instead of unbalanced RCA between every gear. It helps, but not all.

Apparently there is ground loop. This cannot be broken by balanced driver.

I wanted to use signal transformer, to break ground loop. Looking in the net, Jensens are expensive, because will use quite some of them. Tried the cheap Asian made, the result are not good. The highs are missing.

So wanted to DIY this like transformer. Anyone can help how to built the good one? What will be the core size, winding numbers, wire sizes etc?

Please dont direct me "just buy Jensen !"

[TonyTecson]

hi,

tell me what materials (core size/types) magnet wires are available at your end...

likewise, you desired input/output impedance, or the signal levels that your diy traffo will see, then we can try....


tony

[djk]

You are wasting your time.

You only need a transformer at the receiving end.

Buy EDCOR.

http://www.edcorusa.com/sound/matchers/s2s10k_10k.htm

[lumanauw]

Hi, DJK,

DIY line transformer will not give satisfactory result? Is it have to use certain core material?

[JonMarsh]

I used to design and build moving coil step up transformers in the 70's.

A good transformer is a carefully crafted series of tradeoffs. The reason that Jensen's are so widely known and respected is that they are very well designed and manufactured using very good materials, and overall the pricing is reasonable considering what you get.

As DJK so diplomatically pointed out, you don't need transformers at both ends. You can use just a line bridging/receiving transformer at the input side of your equipment, and drive it from either an active balanced or single ended source. With good transformer design, you can expect to get over 60-80 dB of common mode rejection at line frequencies.


The primary design goal for any transformer is to have as high an effective magnetizing inductance on the primary as possible (this minimizes the source impedance reuqired to drive it and sets the LF extension corner frequency and low frequency distortion), while at the same time minimizing the uncoupled leakage inductance between windings and lamination hysterisis effects, as the latter are the main mechanisms for HF roll off in the transformer. Typical lamination materials for high performance audio transformers are made of a moly permalloy alloy with very thin lamination thickness. I don't have the source catalogs any more that I used in the 70's; these are specialized items.

This is a juggling act; if you want really "good" group delay numbers in the low frequencies, you need to extend the LF response to below 2 Hz, but this makes getting good HF response more difficult. A reasonable minimum target for a good sounding transformer is at least 3-4 Hz -3 dB point on the bottom end, and 80-100 kHz on the top end.

My favorite line bridging/input transformer is the Jensen JT-6110K-B. This part has good bandwith driven from a 600 ohm source (-3 dB at 0.15 Hz and 120 kHz), and 130 dB of CMRR at 60 Hz.

Now, if you have very low impedance line driver circuits, you can "cut corners" a bit and use/design a line transformer for a low impedance drives; this makes things easier, as long as you have the necessary driver. I did this in one electronics crossover for which I desired balanced outputs with true isolation (non active balanced), and used the JT11SS. This part has much lower primary magnetizing inductance, and is really happiest being driven from a 75 ohm or lower impedance source (TRUE 75 ohm power bandwidth capability, not just LF impedance from feedback loop!). In this case, AD715 video buffers, which are in a high power SIP package (kind of like the gainclone chips) were used.

Distortion and bandwidth are dependent on source impedance- below 50 ohms is desirable. For this kind of transformer, the tradeoff is that lower primary inductance lowers the net leakage inductance, which improves the bandwidth in the top end substantially (to beyond 450 kHz). The tradeoff is needing a brute of a line driver. Can't really drive units like this well with DIP8 opamps. But for DIY, that shouldn't be a big deal...

~Jon

[Eva]

I think that having electronically balanced inputs or floating electronically balanced outputs and using shielded twisted pair wire is enough to solve the hum problem. If one side is unbalanced, impedance balancing resistors would be required. See Douglas Self site [search in google] to understand' impedance balancing' concept

[djk]

" I think that having electronically balanced inputs or floating electronically balanced outputs and using shielded twisted pair wire is enough to solve the hum problem. If one side is unbalanced, impedance balancing resistors would be required. See Douglas Self site [search in google] to understand' impedance balancing' concept"

Well neither of you understand where the hum is coming from.

To eliminate the hum requires galvanic isolation, ie: a transformer.

Very good transformer tutorial at Jensen and Rowland sites.

[ChocoHolic]

I think Eva's proposal would also be fine.
The hum due to ground loops is caused because normally we use
ground as the signal reference. Then we connect our system grounds
at different points to earth. As a result the ground at one end is different from the ground from the other end (typical some millivolts at 50Hz/60Hz), which means our signal references are different... !
If you would use a signal that is not related to ground/earth, but only the difference between two non grounded wires is carrying the information, and use symmetrical/differential input, then the 50Hz/60Hz currents in the ground loops will not directly add hum to signal.
But it only works if you do not mess up the signals with wrong ground connections again... So a twisted pair would not be bad, if you want to screen such twisted pair then you should connect the screen only at one end to ground other wise the different ground levels may cause a current in the screen, which might couple into the signal by induction...
You will only get lucky if you are exactly knowing what you are doing.

I think, a galvanic isolating transformer at the input is much easier to handle. It typically also works if nobody knows what he is doing.

This is not meant in a negative way.
In fact I love transformers, because I love rugged and forgiving systems.

There are also special cables available, which include a transformer
that does not isolate but simply adds exactly the difference of the ground signal also to the signal itself ... We call them "Mantelstromfilter" in Germany...
But they should be popular in various regions, not only Germay.
http://www.hifi-regler.de/hifi/mante...57ce63977f7adb
They work, but are not really high end...


Cheers
Markus

[Eva]

Balanced signal transmission is inmune to ground loops. With properly designed equipment the ground loop currents may even be left flowing freely through the screen of the twisted pair [connected at both ends] and nothing will happen. For better CMRR, impedance balancing is also required

Note: A 'Mantelstromfilter' appears to be nothing but a 'common mode filter'

[Nelson Pass]

Originally posted by Eva
I think that having electronically balanced inputs or floating electronically balanced outputs and using shielded twisted pair wire is enough to solve the hum problem. If one side is unbalanced, impedance balancing resistors would be required. See Douglas Self site [search in google] to understand' impedance balancing' concept

You can also find a discussion in the A75 article(s) on
www.passdiy.com. This works quite well, but there's nothing
like a Jensen or comparable transformer for the finest in
noise rejection and avoiding ground loops.