As Juan Fahey mentioned, winding audio transformers working in the full audio band is not that easy. Until now you focused on the low frequency end where core saturation is the dominant issue.
With your EP core and more than 2500 turns the resulting stray inductance cannot be neglected - and this affects the upper frequency end. Additionally you will have to consider the drive and load impedances in this game - they might be very restricted for a flat frequency response.
Ferrite is a poor material for audio transformers due to its low saturation level. This forces you to use a extra high number of turns to avoid saturation. With the result of high copper resistance and high stray inductance.
You will find that professional audio transformers are fitted with laminated iron.
Some years ago I have designed an 1:1 audio transformer with an EP13-core. I did this because at that time ready made transformers went out of stock. And it was a challenge to choose ferrite!
500 turns each winding an the core with highest permeability money can buy. Passband of interest was 300Hz~4kHz: An analogue telephone interface. And believe me it took enormous efforts to create an equalization network to correct line impedance matching.
Anyway, a bigger core helps reducing these parasitics, you may end at some bigger RM or pot cores.
With your EP core and more than 2500 turns the resulting stray inductance cannot be neglected - and this affects the upper frequency end. Additionally you will have to consider the drive and load impedances in this game - they might be very restricted for a flat frequency response.
Ferrite is a poor material for audio transformers due to its low saturation level. This forces you to use a extra high number of turns to avoid saturation. With the result of high copper resistance and high stray inductance.
You will find that professional audio transformers are fitted with laminated iron.
Some years ago I have designed an 1:1 audio transformer with an EP13-core. I did this because at that time ready made transformers went out of stock. And it was a challenge to choose ferrite!
500 turns each winding an the core with highest permeability money can buy. Passband of interest was 300Hz~4kHz: An analogue telephone interface. And believe me it took enormous efforts to create an equalization network to correct line impedance matching.
Anyway, a bigger core helps reducing these parasitics, you may end at some bigger RM or pot cores.
Last edited:
Yes, in US. In Europe I pay less than 20 Euro for a high quality mu-metall shielded audio transformer, made by pikatron, available at bürklin.
Yes, in a transformer this 'stray inductance' is called the 'leakage inductance'. It appears in series with the load, which in this case is a fairly high impedance - the input resistance of the TDA8932 in balanced mode which is 50k. The leakage inductance would have to contribute a substantial fraction of 50k load resistance to make an impact on the FR.
The source impedance from an audio source is generally fairly low, unless the source is a valve preamp. Typically under 500ohms. As already pointed out, the load impedance is high. I have made measurements of the FR of my transformers to well beyond the audio band, driven from my signal generator (which has 50R output impedance) - no appreciable roll-off was seen.
In comparison with steel or mu-metal I agree, ferrite is poorer in terms of its maximum flux. But that's only one parameter of relevance to transformer design. Those other materials don't suit themselves to being DIY'ed which is the point of this thread. For a signal transformer the higher resistance of the copper is still not significant with such a high load impedance.
Going for the highest permeability material might not be a wise choice given there's normally a tradeoff between flux handling and permeability. Higher mu means lower peak flux typically. So why the choice of highest mu here? EP13 also is a very small core (Amin = 15mm^2) so was size a primary constraint?
Did you have to drive a 600 ohm load? And have 600 ohm source impedance? In which case I can see you'd have much more of a problem with leakage inductance than in my case.
A bigger core has other issues which might well be detrimental in an audio application, as in all engineering there are tradeoffs.
When I've watched the output level of one of my transformers vs frequency with only my AC voltmeter on the secondary what I notice is a rising response beyond the audio band, not a roll-off. I'm not sure of the input impedance of the voltmeter but most likely an RC snubber on the secondary would be a good way to bring the overall response back to flatness. The trafo wants to see some kind of resistive load to dampen its inherent HF resonance.
Last edited:
I'm already getting satisfactory results. I agree mu metal can handle higher flux, but its not so amenable to DIY.
Incidentally for some more background information about transformers, this set of white papers from Lundahl is very useful. Especially chapter 6 which shows equivalent circuits for transformers including leakage inductance.
Last edited:
Tool for designing transformers:
http://www.diyaudio.com/forums/soli...imate-fidelity-amplifier-778.html#post4758987
http://www.diyaudio.com/forums/soli...imate-fidelity-amplifier-778.html#post4758987
Sorry, just discovered this - here's the link
As I wrote - Bürklin (shop) pikatron (brand):
https://www.buerklin.com/en/search?text=pikatron
(scroll down page)
Last edited:
Note that's for power transformers which are far more complex to design than audio signal transformers. As such the choice of cores available is limited to those suited for power.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.