The Mad Science Challange -DIY MIC IPT

[abbey road d enfer]

Hi thanx 5v333… I have signal generator but no scope, Im thinking about buying a scope so I can measure those high freq.

Are you familiar with oscilloscopes?
If not, I would recommend you get a second-hand CRT 'scope, rather than one of the new digital ones.
Second-hand CRT scopes can be found for very little money. The main issue is they are big and heavy compared to new ones.

 

[solkatten]

Om not that familiar with scopes but I was looking for a second hand one… I look for a CRT

 

[solkatten]

I took some measurements on my home brewed trafos. THD+N measured at low level ca. -35dB primary side.

Red is the focus rite as reference for freq. response.
Black is a ca 1:4,5 toroid THD 0,013% 50Hz (smaller core pri inductance 17H)
Green is LL1538 THD 0.018% 50Hz (pri inductande: 20H)
Blue is 1:7 toroid THD 0.0088% 50Hz (bigger core pri inductance: 10H)

The reference IPT: https://www.lundahltransformers.com/wp-content/uploads/datasheets/1538_8xl.pdf

 

[gyraf]

1:8 @200R Source?

..nicelygood..

/Jakob E.

 

[solkatten]

You can´t beat the measurment equipment...

Sound card is -1dB at 35KHz
1:4,5 is down -1dB at 35KHz (as flat as possible)
LL1538 is -1dB at 32KHz
1:7 is -1db at 22KHZ

 

[solkatten]

1:8 @200R Source?

..nicelygood..

/Jakob E.

Its 1:7 with a 200 ohm source. The electrostatic screen between primary and secondary was the game changer for high freq performance...

 

[abbey road d enfer]

I took some measurements on my home brewed trafos.
Red is the focus rite as reference.
Black is a ca 1:4,5 toroid THD 0,013% 50Hz
Green is LL1538 THD 0.018% 50Hz
Blue is 1:7 toroid THD 0.0088% 50Hz

Turns ratio determine fred response and they all perform well...

If I read correctly the graph, there is a loading issue, since the 1:8 (1:7?) xfmr brings only 4.4dB of gain and the 1:6 (1:4.5?) only 1.5dB.
Can you check the primary voltage response?

 

[solkatten]

If I read correctly the graph, there is a loading issue, since the 1:8 (1:7?) xfmr brings only 4.4dB of gain and the 1:6 (1:4.5?) only 1.5dB.
Can you check the primary voltage response?

hey I updated the turns ratio in the new picture. I corrected the turns ratio with LL1538 as a reference 1:5. 1:6 was me misreading the measurment. The 1:4,5 was suppose to be 1:4 but because the secondary is laid on top the diameter is increasing. The 1:8 turned out 1:7 maybe an error in my calculation in turns per section.
They all got the same signal level on their primary.

So my toroid steps up 16dB 1:6,5ish, I can not claim 1:8....

 

[abbey road d enfer]

They all got the same signal level on their primary

So where's the signal loss?

 

[solkatten]

So where's the signal loss?

The green line is LL1538 its steps up the signal 14dB, my toroid steps up signal 16 dB and my other toroid steps up the signal the signal 13 dB. Im correcting the turns ratio according to the measurement

 

[solkatten]

The confusion was me using the 1:4,5 as reference then I though I used LL1537 as reference,,, sorry

The sound card in red is only there as reference of freq. response, not as dB reference...

 

[abbey road d enfer]

The sound card in red is only there as reference of freq. response, not as dB reference...

Ah OK.

 

[solkatten]

I will have one last shoot at a good performing 1:8 IPT. Goal: Flat response beyond 20KHz with 200ohm source....

First a singel layer 0,05mm 1689T (sec winding)
Electrostatic screem between pri and sec.
Then a singel layer 0,4mm 210T (pri winding)


The compromise will be slightly lower primary inductance = 4H25. The singel layer aproach will keep both primary and secondary within perfect symmetri and capacitance and the voltage difference between turns and windings will be keept at a minimum. The large core has low distortion (half the THD at half the primary inductance compared to LL1538) and I dont need the sub sonics.

 

[solkatten]

Otherwise I think I am satisfied with my 1:4-5 design, they are comparable or better than LL1538. In regards of THD and high freq. response. Because inductance is flat over the freq. range (in vitroperm 500F) it don´t have that inductance boost in the low freq. range that other core materials have. So LL1538 wins the bass compartment. Toroidal core pros are high cmrr and low core losses.

 

[solkatten]

A singel layer secondary with a tiny wire gauge I think this one will be a killer
And Merry Christmas Everybody

 

[abbey road d enfer]

The compromise will be slightly lower primary inductance = 4H25. The singel layer aproach will keep both primary and secondary within perfect symmetri and capacitance and the voltage difference between turns and windings will be keept at a minimum. The large core has low distortion (half the THD at half the primary inductance compared to LL1538) and I dont need the sub sonics.

It would be interesting to measure the inductanc at LF; actually measuring the impedance at several frequencies and levels would be very enlightening.
Apart from the large size, I can see only advantages at using a large core.
What do you use for the faraday shield?

 

[solkatten]

It would be interesting to measure the inductanc at LF; actually measuring the impedance at several frequencies and levels would be very enlightening.
Apart from the large size, I can see only advantages at using a large core.
What do you use for the faraday shield?

Copper tape on top of a plastic tape… see post 32.

 

[gyraf]

Apart from the large size, I can see only advantages at using a large core.

I've avoided larger sizes for two reasons until now:

- I have a not-yet-quantitatively-substantiated feeling that barkhausen dist (=very-low-level-resolution) gets larger with bigger cores. At least something less than desirable happens in the shadows / room tails when going bigger.?

- Increased length-of-wire-on-core results in higher inter-winding capacitance, higher DC resistance, or both..?

I may be wrong on both accounts though, I'm still going through the possible parameters one by one whenever there's time for experiments (and not much spare time recently, plenty of customers in the commercial department - which is a very good thing)

And yes, I caved in and ordered a winding machine so I can get through iterations a bit quicker. It won't be here until April though..

/Jakob E.

 

[solkatten]

Hi about barkhausen dist I have no clue. But the inductance per wire lenght actually stays the same between my to cores despite the size difference

 

[abbey road d enfer]

- I have a not-yet-quantitatively-substantiated feeling that barkhausen dist (=very-low-level-resolution) gets larger with bigger cores. At least something less than desirable happens in the shadows / room tails when going bigger.?

I haven't seen much info about how Barkhausen changes with size. It would seem logical that the lesser the induction, the higher the Barkhausen steps in relative value, but actually the reducing slope at the beginning of the B-H curve may compensate that.
Just speculating here, since I've never been annoyed in practice by Barkhausen.

- Increased length-of-wire-on-core results in higher inter-winding capacitance, higher DC resistance, or both..?

Higher Al results in less turns. There's definitely increasing returns in increasing size. Number of turns decrease as a function of volume, hence cube of linear dimensions. Length of wire is proportional to square of linear dimensions. Actually wire size can increase as much as dimensions, resulting in less DCR, or a compromise with slightlt lower DCR and stable capacitance.
Again this is all theoretical.

And yes, I caved in and ordered a winding machine so I can get through iterations a bit quicker. It won't be here until April though..

What machine have you ordered?