Valve mic preamp design incoherent rambling

[Tubetec]

Yes REW also works with Mac os .

I have the XJW01 LCR off ebay , great for measuring winding capacitances and low resistances fairly accurately , 100hz , 7.8khz and 10khz test frequency ,where the inductance at 50 hz in transformers is the usual reference point.
Theres a few good tutorials written here about measuring inductance with a sig gen and multimeter . With small signal transformers something like REW and your soundcard output should be able to generate the test tones adequately .

 

[solkatten]

Im actually going to get a better LC meter so I can measure at diffrent freq. So if I lay the primary 0,15 mm = 360T = ca. 7H. would be more fair.

Yes REW also works with Mac os .

I have the XJW01 LCR off ebay , great for measuring winding capacitances and low resistances fairly accurately , 100hz , 7.8khz and 10khz test frequency ,where the inductance at 50 hz in transformers is the usual reference point.
Theres a few good tutorials written here about measuring inductance with a sig gen and multimeter . With small signal transformers something like REW and your soundcard output should be able to generate the test tones adequately .

I actually got on old signal generator with a 600 Ohm output, hook up that to the primary of ITP and have a secondary load resistor to reflect 6K to primary. Measure freq. Response with a DVM on the secondary. Resonance peak at the top and roll off at the low.

 

[Winston OBoogie]

Response of DVM's isn't that great as you go higher in frequency. If you have a scope to look at the voltage, you'll get a better idea of what's going on up where the resonance is. With a 2 channel scope, you can monitor input and output to make sure your IP voltage is constant as you step through frequencies.

With a mic input transformer, it's important to have the source impedance be the same as a typical mic impedance for checking frequency response so, a suitable pad from your 600 ohm generator to 150 - 200 ohms is the way to go.

Edit: you'll still need a suitable pad if you were using a USB interface and a computer with REW.

Easy stuff to whip together for an industrious chap like yourself

 

[solkatten]

Response of DVM's isn't that great as you go higher in frequency. If you have a scope to look at the voltage, you'll get a better idea of what's going on up where the resonance is. With a 2 channel scope, you can monitor input and output to make sure your IP voltage is constant as you step through frequencies.

With a mic input transformer, it's important to have the source impedance be the same as a typical mic impedance for checking frequency response so, a suitable pad from your 600 ohm generator to 150 - 200 ohms is the way to go.

Edit: you'll still need a suitable pad if you were using a USB interface and a computer with REW.

Easy stuff to whip together for an industrious chap like yourself

Yep I just discovered that it is hard to measure hi freq in this setup I measured a LL 1538 and it sucked as well. But the -3dB was 10 hz with a 600 Ohm source I pad the source to 200 Ohm.
The DI input on my sound card should high enough impedance not mess up the measurement

 

[solkatten]

Response of DVM's isn't that great as you go higher in frequency. If you have a scope to look at the voltage, you'll get a better idea of what's going on up where the resonance is. With a 2 channel scope, you can monitor input and output to make sure your IP voltage is constant as you step through frequencies.

With a mic input transformer, it's important to have the source impedance be the same as a typical mic impedance for checking frequency response so, a suitable pad from your 600 ohm generator to 150 - 200 ohms is the way to go.

Edit: you'll still need a suitable pad if you were using a USB interface and a computer with REW.

Easy stuff to whip together for an industrious chap like yourself

Response of DVM's isn't that great as you go higher in frequency. If you have a scope to look at the voltage, you'll get a better idea of what's going on up where the resonance is. With a 2 channel scope, you can monitor input and output to make sure your IP voltage is constant as you step through frequencies.

With a mic input transformer, it's important to have the source impedance be the same as a typical mic impedance for checking frequency response so, a suitable pad from your 600 ohm generator to 150 - 200 ohms is the way to go.

Edit: you'll still need a suitable pad if you were using a USB interface and a computer with REW.

Easy stuff to whip together for an industrious chap like yourself

Im going to try measure with REW with correct impedance

 

[5v333]

The input of your ADC prob has a much lower Z than 60k.

Also make sure that you have the transformer in the right configuration.

 

[solkatten]

The input of your ADC prob has a much lower Z than 60k.

Also make sure that you have the transformer in the right configuration.

Yep the 60k is the line input impedance of the sound card… the 220r source impedance i set by a -20db pad…

 

[solkatten]

Hi here here are 2 measurments from REW. First one is Lundahl 1538 for reference. Second is my hand wound toroid. Both have flatter response than my soundcard... Lundahl has higher primary inductance and wins in the THD department.

Its an old sound card audio box so it does contribute to the THD, my new focusrite dont want cooperate...

My toroid:
Turns ratio: 1:6,5

Primary inductance: 7H4 at 140Hz
Interwinding capacitance (measured by short circiut primary and secondary) 267p
Capacitance across secondary winding: 278p

Primary winding a singel layer 0,15 mm
Secondary on top: semi progressiv winding 0,1 mm

 

[gyraf]

..try to measure with a resistor e.g. 200 or 600 Ohms in series with the primary - to stand-in for real-life source impedance..?

/Jakob E.

 

[solkatten]

..try to measure with a resistor e.g. 200 or 600 Ohms in series with the primary - to stand-in for real-life source impedance..?

/Jakob E.

After the sound card output I have -20db pad 1k x2 and 220r. So primary sees 220r. The secondary sees the high impedance of the DI input

 

[solkatten]

(Error in measurment true source impedance was 1K3)This measurment is with a 3k9 source resistor to provoce some roll off. The -3db is at 38Hz. That means that the primary inductance increases at low Hz to ca 16H...
The high -3db is at 7KHz, -3db for LL1538 at 3k9 source is 14KHz.

 

[solkatten]

Lessons learned. After feedback on the forum that the singel layer secondary design wasnt optimum, I wound a new IPT 1:5. First a singel layer 0,1 mm (secondary 1), than a layer of .25mm (primary) then on top a section semi progressive 0,1mm (secondary 2). Primary inductance is low at 1H9 so it cant take massive signals. But when loaded with 3k9 source it rolled of earlier -3db at 5KHz, which I didnt expected. Interwinding capacitance is higher at ca 350pf and capacitance across the secondary 430pf.

 

[gyraf]

aah yes, the whole art is the balancing of all these parameters - you seem to be on the right track though!

/Jakob E.

 

[solkatten]

aah yes, the whole art is the balancing of all these parameters - you seem to be on the right track though!

/Jakob E.

Thanx for the encouragement
That is great because it is one of few ways it is possible to hand wound toroids And its pretty easy to…

 

[solkatten]

A bit of comparing data:
I measured interwinding capacitance (IW) and capacitance across secondary (SW) with 2 diffrent capacitance meter an got quite diffrent results... I have no means of making calibrated measures, so the results dont have any absolute meaning, but can be a relative indicator of performance...

LL538:
IW: 437p and 250p
SW: 2n2 and 340p

Toroid (2 layer)1:6,5:
IW: 560p and 260p
SW: 680p and 320p

Toroid (3 layer) 1:5
IW: 1n1 and 265p
SW: 4n4 and 390p

If you were to extrapolate from this figures my 2 layer toroid should be the winner but its not. The LL1538 has better treble response. The LL1538 has very orderly and symetrically wound secondary and has elctrostatic screen between pri and sec. All that contributes to hi freq response.

The toroidal transformer has a perfect symmetrical primary which provides good cmrr. It has no electrostatic screen and the secondary cant be hand wound in perfect symmetrical order (only a progressive toroidal winding machine could do that.. expensive)

The one I could improve upon is to use an elctrostatic screen between pri and sec. Its tricky but can be done...

Leakage inductance and such arent of a big concern in toroidal design...

Do you guys have any ideas how to improve the IPT? Whitin the boundarys of a hand made design?

 

[solkatten]

A way forward... I have ordered a bigger core in vitroperm 500F-series 45mm AL=88uH. With a singel layer primary .2mm I get 17H. I will wind with shuttle by hand, but I will build a guide for the wire, so I can lay down .1 mm wire in progressive maner... I will also have a electrostatic sheild between pri and sec...

 

[gyraf]

..interesting range of experiments, thanks for sharing with us..

notes:
1 - with a larger core you get not only increased AL, but also squarely-increased length-of-each-turn - and thus "capacitance-per-henry"
2 - remember that the turns/layers of your electrostatic screen must be isolated from each other, so it does not constitute a shorted turn. I've sandwiched alufoil between two layers of tape for this - with a wire attached to one end only

/Jakob E.

 

[solkatten]

..interesting range of experiments, thanks for sharing with us..

notes:
1 - with a larger core you get not only increased AL, but also squarely-increased length-of-each-turn - and thus "capacitance-per-henry"
2 - remember that the turns/layers of your electrostatic screen must be isolated from each other, so it does not constitute a shorted turn. I've sandwiched alufoil between two layers of tape for this - with a wire attached to one end only

/Jakob E.

Thanks for the advice I have ordered singel layer conductive copper tejp 6mm. So I can wind it around the toroid without making a complete turn. For insulation I use teflon tape which I pre- wind on a shuttle. Yepp I know there are hi freq trade offs with larger core. But it will be easier to work on a larger core, and you can use manageble wire gauge.. My hope is that a larger core and primary inductance and more orderly wound secondary + electrostatic screen will rival THD and freq response of the Lundahl...

When comparing freq response against LL1538 one have to remember that transformer does +-0,3 db 10Hz to 100KHz with a 200r source.

 

[solkatten]

Actually the inductance per lenght of wire pretty much stays the same between the 2 cores...

Small core 30x20x10 Al=56 30mm/turn
Large core 45x30x15 AL=88 45mm/turn

 

[solkatten]

The bigger cores are on back order so they want come soon

But from the 3k9 source measurement one can extrapolate the freq response of my toroid with a 220r source impedance, I know the high end behavior is more complex than a low pass filter, besides capacitance it has leakage inductance etc... it has self resonanse I cant see.
(Errror in measurment) In the low end the -3db is at 2Hz because inductance rises at low Hz to 16 H.
At the high end the measurement indicates -3db roll off at 120KHz

Would it be correct to calculate self resonance freq. by having total capacitance from freq plot: 5n8 + measured leakage inductance?