OPEN SOURCE DIY Mic Project - ORS 87 - Stripped Down u87

[AdamFrisen]

I will attach the project I used in QSPICE (for all us Windows users only I guess):

I think I used this model for 2N3819 (https://groups.google.com/g/sci.electronics.components/c/PGImbJDq3bw)

.model J2N3819 NJF(Beta=1.57m Lambda=4m Vto=-1.954
+ Cgd=1.5p M=.3821 Pb=1 Fc=.5 Cgs=1.5p Is=173.3f
+ N=1 Xti=3 Kf=37.24E-18 Af=1)

 

[OneRoomStudio]

I think you're forgetting that signal would also be hitting the rear diaphragm (in opposite polarity) in the original.

 

[AdamFrisen]

I think you're forgetting that signal would also be hitting the rear diaphragm (in opposite polarity) in the original.

I don't know how to model that . I started to use simulating programs yesterday.

I just put same signal to the messeingang.

 

[Khron]

Darn yet another capacitor I don't have...

Two 10pF in series?

 

[pasarski]

Darn yet another capacitor I don't have...

Roll your own

 

[OneRoomStudio]

I simulated 3 circuits, U87 original, U87 ORS without C1 and with C1 = 10pF.

View attachment 134752

And I got this result:

View attachment 134753

If (big if) I draw the schematics properly, it looks as the U87 ORS with C1 = 10pF looks more similar to U87 in the response than without?

I didn't know it was so easy (and fun) to simulate analog circuits (even if I might have done it wrong).

This is like when I discovered the benefits of writing unit tests in software for the first time.

As both @kingkorg and I have been pointing out - you're better off adjusting C5/C6 to match the response. Skip C1 and try C6 = 100pF and C5 = 100nF.

 

[AdamFrisen]

Two 10pF in series?

Roll your own

 

[AdamFrisen]

As both @kingkorg and I have been pointing out - you're better off adjusting C5/C6 to match the response. Skip C1 and try C6 = 100pF and C5 = 100nF.

Thanks. I don't have so many different capacitors except those suggested earlier like 160pF, 220pF, so I will try to use what I have at the moment and see where I land (I do have 3.3 pF from Khron's BOM for testing out NFB).

I do like that THD went down 10x with the feedback capacitor. In simulation connecting C1 directly from gate to drain instead as in schematics, gave the same result as other pointed out it would.

 

[Khron]

I do like that THD went down 10x with the feedback capacitor.

And that's correlated with how much decrease in signal gain?

 

[kingkorg]

And that's correlated with how much decrease in signal gain?

And noise gain?

 

[AdamFrisen]

And that's correlated with how much decrease in signal gain?

I have only tried 3.3pF on the flat circuit so far. With same signal in of 200mV RMS to messeingang, I got following data. The cursor is at about 1KHz.

No C1:
Out signal: -12.8 dBu
THD (H2.9): 0.99%

With C1=3.3pF, soldered directly to the gate and drain:
Out signal: -27.3 dBu
THD (H2.9): 0.091%


With no C1:



With C1 = 3.3pF:

 

[OneRoomStudio]

14.5dB down is around 5.5x less signal. The problem is that the noise stays pretty much the same (as you can see in the charts - around -110dB), so your SNR is much worse. The THD is better, but only because there’s less signal gain. I’d take slightly higher THD on the transients over noise any day.

 

[cello4u]

The transformer has two sides: Primary (left in the schematic) and Secondary (right). Each side has a start (+) and an end (-).

In order to determine which side is which in your T-13, measure the resistance between the black pair and then the resistance between the yellow pair. The pair with 450 Ohms (approximately) would be Primary. Connect that pair to P+ and P-.

In @Khron's board, the Secondary wires go to S1+ and S2-.

(As for + and -, don't worry unless your mic's output is 180 degrees out of phase -- in that case connect the Secondary wires the other way around [S2- and S1+]. If you don't care about phase at this point, then don't worry and just connect those wires whichever way your prefer.)

 

[cello4u]

The transformer has two sides: Primary (left in the schematic) and Secondary (right). Each side has a start (+) and an end (-).

In order to determine which side is which in your T-13, measure the resistance between the black pair and then the resistance between the yellow pair. The pair with 450 Ohms (approximately) would be Primary. Connect that pair to P+ and P-.

In @Khron's board, the Secondary wires go to S1+ and S2-.

(As for + and -, don't worry unless your mic's output is 180 degrees out of phase -- in that case connect the Secondary wires the other way around [S2- and S1+]. If you don't care about phase at this point, then don't worry and just connect those wires whichever way your prefer.)

 

[AdamFrisen]

As both @kingkorg and I have been pointing out - you're better off adjusting C5/C6 to match the response. Skip C1 and try C6 = 100pF and C5 = 100nF.

Thanks. The deviation from the original frequency response would still be audible even those new values were to be used?

I aligned all the responses to same offset so it is easier to see the difference:

Red line: U87 original

Blue line: U87 ORS with C1 = 5pF, v(out_ors_mod) * 1.05

Green line: U87 ORS without C1, C6=100pF, C5=100nF, v(out_ors) / 3.5




Looking closer to 3KHz - 22KHz:



What would be better values of C5 and C6 to get closer to U87 original?

I am also wondering when +14dB extra gain that U87 ORS gives would be needed compared to U87 original circuitry?
I can't remember I have seen so many posts where SNR of U87 classic has been a major issue, although I agree higher SNR is better.

Either my schematics doesn't reflect the true response of U87 original (then I would much appreciate help to correct it), or it is "good enough" approx. and using C1 in U87 ORS is seems to be a simple way to mimic the response of the original.

The U87A and U87AI doesn't seem to have NFB capacitor? I will try to draw schematic for that in QSPICE .

I see Micro-Cap being praised and mentioned online. Anybody using it and know if it's worth investing time to learn it?

 

[OneRoomStudio]

Just tweak C5/C6 to taste. Maybe 82pf and 150nf.

Also keep in mind that capsule frequency responses vary greatly. Focusing too much on the very linear amplifier frequency response differing slightly from the theoretical original (especially when most caps are +/- 20% in value), is a little silly.

 

[mrchapp]

Every now and then I find my ORS87 doing this every ~0.62641875 seconds:


First time this happened my kid was trying to whistle (he's 7, doesn't know how to) and a puff of air or something really loud and deep made the mic enter that state. It only went away when P48 was turned off. It's a very loud signal. I had to lower the gain there to capture the signal without distortion.

It looks like I can reproduce something similar by getting really close to the capsule (headbasket with windscreen) and saying loudly the word "tres" extending the s:

It doesn't recover from that unless there's silence.

I know what to do: Don't say the word "tres" right next to the capsule and don't blow air into the mic (which is sound advice anyway, pun intended), but I'm just curious on what would be causing this.

 

[rock soderstrom]

I know what to do: Don't say the word "tres" right next to the capsule and don't blow air into the mic (which is sound advice anyway, pun intended), but I'm just curious on what would be causing this.

That sounds like a good strategy, we all do it that way...

Every now and then I find my ORS87 doing this every ~0.62641875 seconds:

I would probably try to isolate the problem first. Change the capsule and see if the problem is still there. If yes, then your impedance converter has problems (my bet), if not the capsule is to blame.

 

[Voyager10]

I don't want to spam this thread, but if anyone's interested I've made some measurements of an ORS87 build here: https://groupdiy.com/threads/comparison-of-jfets-for-mic-applications.86559/post-1164449

This is without the frequency-correction network / messeingang for now, that's coming later.

 

[cello4u]

MP3 to upload later. Now I need to learn how to Bias. This was my third attempt after a couple on another PCB. It's a definite breakthrough for me. Not possible without this thread.