OPEN SOURCE DIY Mic Project - ORS 87 - Stripped Down u87
- Thread starter Wordsushi
- Start date
Help Support GroupDIY Audio Forum:
... And that is exactly how you (most likely ) zapped your JFET.
The gate pin is "in the air", not connected to anything..? Any particular reason for that? The 1Gohm resistor that's supposed to be connected to it, is soldered on the board, i see.
Why would you have needed to measure the capsule bias voltage anyway, though?
You're right, I'll re-solder that
It has very little gain Apollo Twin 10dB full volume for headphones DT 770 and I sound minimal it sounds muffled and very quiet compared to a brown Phanthera Microphone In the UAD display I come when I speak normally not about -27 dB the brown phanther test is at -13dB
It has very little gain Apollo Twin 10dB full volume for headphones DT 770 and I sound minimal it sounds muffled and very quiet compared to a brown Phanthera Microphone In the UAD display I come when I speak normally not about -27 dB the brown phanther test is at -13dB
STOP PROBING ANYTHING ON THE JFET GATE!
It might help if you also read and understood other people's advice too:
Last edited:
TIDStudios
New member
@RuudNL So are you just swapping out the c104 47nf cap for a 100nf cap, and changed the 160nf to 180nf? Is that all?
Voyager10
Well-known member
As an aid to troubleshooting, here's some information about the ORS87-Plus testbed I built.
For testing, the U87's 'calibration input' arrangement was rigged up by replacing R15 with 560R + 6K8 in series, and the input wired to the midpoint. (I understand the V1.1 PCB has space for 2 resistors). I didn't implement a pad or bass-cut switch. Feedback capacitor C4 was 2.7pF, made by putting 5pF and 6pF caps in series.
To simulate a single cardioid K87-style capsule, there's a 47pF capacitor (Ccap) connected between the FD and BP inputs (the Neumann test procedure includes a fixture which does exactly this).
The BF256B JFET was chosen after a shootout (here) between various types: it's nearly the best by various measures, and cheap and readily available. Biasing the FET was done by adjusting trimmer R7 for minimum THD at the output, when fed with a 100mV, 1KHz sine wave at the calibration input. Values in blue are DC measurements at the bias point, and in red are the measured AC values.
The input stage looks like this:
I also did a frequency response and distortion plot using REW. This is with a 100mV RMS input level.
We're getting a moderate 3-4dB dropoff at 15KHz, and a stronger rolloff after that. Obviously some of the distortion is coming from the transformer, which is relatively cheap compared to many alternatives.
Schematic
Here's the schematic with the component values used. The changes from the ORS87+ schematic were mostly to follow the original U87/U87A values more closely.For testing, the U87's 'calibration input' arrangement was rigged up by replacing R15 with 560R + 6K8 in series, and the input wired to the midpoint. (I understand the V1.1 PCB has space for 2 resistors). I didn't implement a pad or bass-cut switch. Feedback capacitor C4 was 2.7pF, made by putting 5pF and 6pF caps in series.
To simulate a single cardioid K87-style capsule, there's a 47pF capacitor (Ccap) connected between the FD and BP inputs (the Neumann test procedure includes a fixture which does exactly this).
The BF256B JFET was chosen after a shootout (here) between various types: it's nearly the best by various measures, and cheap and readily available. Biasing the FET was done by adjusting trimmer R7 for minimum THD at the output, when fed with a 100mV, 1KHz sine wave at the calibration input. Values in blue are DC measurements at the bias point, and in red are the measured AC values.
The input stage looks like this:
Results
Results are pretty close to the reported values for the U87A:| Measurement | Value |
|---|---|
| Overall gain, CAL to Vout, 1KHz (measured at 100mV in) | -1.6dB (0.83x) |
| JFET open-loop gain (1KHz) | +19dB (9.0x) |
| THD, 100mV RMS in, 1KHz | 0.013% |
| Max input voltage. THD < 0.5%, 1KHz (RMS) | 530mV |
| Drain current at optimum bias | 0.38mA |
| Polarization voltage | 45.0V |
I also did a frequency response and distortion plot using REW. This is with a 100mV RMS input level.
We're getting a moderate 3-4dB dropoff at 15KHz, and a stronger rolloff after that. Obviously some of the distortion is coming from the transformer, which is relatively cheap compared to many alternatives.
Attachments
peri valencio
New member
Hello everyone, my contribution here is to read everything very attentively and countless times
. I already have my v1.1 working and I’m very happy, but I’m still testing different components and capsules. Thank you all
. I see a lot of discussion about which types of capacitors to use for coupling; there are so many types of capacitors but few insights on the final results. I found this paper from TI, and it seems to align with what has been written here by the more experienced members.
https://www.ti.com/lit/an/slyt796a/slyt796a.pdf?ts=1729861324029
. I already have my v1.1 working and I’m very happy, but I’m still testing different components and capsules. Thank you all. I see a lot of discussion about which types of capacitors to use for coupling; there are so many types of capacitors but few insights on the final results. I found this paper from TI, and it seems to align with what has been written here by the more experienced members.https://www.ti.com/lit/an/slyt796a/slyt796a.pdf?ts=1729861324029
joulupukki
Well-known member
What effect would a lower-value resistor have? Right now I converted it from 56kohm to 47kohm now with the 33V Zener. With the 33V Zener and 47K resistor I'm not noticing that much difference compared with the 24V Zener & 56K resistor.
I'm still trying to understand some peculiarities of all the U87 variants and understand the design choices that the Neumann engineers made. I hope you guys don't mind that I will (ab)use this thread to educate myself. But of course, I hope I can contribute as well.
Simulating a circuit in LTspice always helps me a lot in understanding a circuit. So that's what I did. I applied some simplifications to Voayger10's circuit and dropped it in LTspice. Picture attached. I was hoping, of course, to get results that would match the measured data on the real circuit. With the components displayed, I obtained a 0.38mA bias current, resulting in a Drain voltage of 11.4V. So far, so good. Ran a sim and checked the gain and high-end roll-off. As you can see from the resulting AC Analysis plot, mid-band gain equals 19 dB and at 15 kHz, the drop-off reached 3.5 dB. Spot on. Or not...?
@Voyager10 : In your table, you mention an Open Loop gain of 19dB. Which is very close to my simulation, but in my view that would be the Closed Loop gain. My definition of Open Loop gain of a circuit would be the gain with all feedback removed. Feedback is through C4 and the feedback circuit from C8 back to the capsule. When I change C8 and C4 to 0.01pF, effectively removing feedback, I obtain an Open Loop gain of 27.6 dB. See last AC Analysis plot attached. So my question is: do we have a different understanding about what the concept of Open Loop entails? Or is my simulation completely off?
Now, getting to @joulupukki 's question: assuming my simulations are right, then changing the zener to 24V, R2 to 47k and rea-adjusting the bias resistor R7 to 10k to obtain 11.2V again at the Drain, the sim shows a circuit gain gain of 18.5dB. So it does very little to the gain, as you already observed. That is easily understood when you realize that only the Open Loop gain is affected by these changes. Open Loop gain is particularly high in this circuit, so it will have some effect on Closed Loop gain, but not a lot. Open Loop gain with 56k/24V is 26.7 dB, so a drop-off of 0.9 dB, resulting in a 0.5 dB Closed Loop gain.
If one wants more Closed Loop gain, remove C4. But I don't see this capacitor in every U87xx incarnation, so maybe yours doesn't have it. Without C4, Closed Loop gain is very close to Open Loop gain, the odds that you will get more gain from this circuit are negligible. I think the best way to obtain a higher sensitivity is by raising the capsule voltage by using a polarization voltage generator circuit. With 60V, you'd have 2.5dB more gain and at 80V, +5dB. But the question is whether your capsule (Arienne's K87, AFAIK) can handle such voltages. Does anyone know?
Jan
Simulating a circuit in LTspice always helps me a lot in understanding a circuit. So that's what I did. I applied some simplifications to Voayger10's circuit and dropped it in LTspice. Picture attached. I was hoping, of course, to get results that would match the measured data on the real circuit. With the components displayed, I obtained a 0.38mA bias current, resulting in a Drain voltage of 11.4V. So far, so good. Ran a sim and checked the gain and high-end roll-off. As you can see from the resulting AC Analysis plot, mid-band gain equals 19 dB and at 15 kHz, the drop-off reached 3.5 dB. Spot on. Or not...?
@Voyager10 : In your table, you mention an Open Loop gain of 19dB. Which is very close to my simulation, but in my view that would be the Closed Loop gain. My definition of Open Loop gain of a circuit would be the gain with all feedback removed. Feedback is through C4 and the feedback circuit from C8 back to the capsule. When I change C8 and C4 to 0.01pF, effectively removing feedback, I obtain an Open Loop gain of 27.6 dB. See last AC Analysis plot attached. So my question is: do we have a different understanding about what the concept of Open Loop entails? Or is my simulation completely off?
Now, getting to @joulupukki 's question: assuming my simulations are right, then changing the zener to 24V, R2 to 47k and rea-adjusting the bias resistor R7 to 10k to obtain 11.2V again at the Drain, the sim shows a circuit gain gain of 18.5dB. So it does very little to the gain, as you already observed. That is easily understood when you realize that only the Open Loop gain is affected by these changes. Open Loop gain is particularly high in this circuit, so it will have some effect on Closed Loop gain, but not a lot. Open Loop gain with 56k/24V is 26.7 dB, so a drop-off of 0.9 dB, resulting in a 0.5 dB Closed Loop gain.
If one wants more Closed Loop gain, remove C4. But I don't see this capacitor in every U87xx incarnation, so maybe yours doesn't have it. Without C4, Closed Loop gain is very close to Open Loop gain, the odds that you will get more gain from this circuit are negligible. I think the best way to obtain a higher sensitivity is by raising the capsule voltage by using a polarization voltage generator circuit. With 60V, you'd have 2.5dB more gain and at 80V, +5dB. But the question is whether your capsule (Arienne's K87, AFAIK) can handle such voltages. Does anyone know?
Jan
Voyager10
Well-known member
My "19dB" figure was with C4 in circuit, i.e. is the gain if the 'outer' feedback loop (C8, C13, R14 etc.) is removed. It will be very close to the closed-loop figure at 1KHz (because the outer feedback loop is largely ineffective at this frequency), so I think this totally agrees with your simulation.
Without C4 I measured a gain of 18x, or +25dB, for the BF256B - that's close enough to the simulation's +27.6dB (25x), especially as my build will have more that 0.01pF of parasitic capacitance...
The theory is that the shape of the closed-loop frequency response curve is affected by the "FET + C4" gain - if you change the FET but adjust C4 to give the same gain, you should get the same frequency response. Higher gain gives more pronounced HF and LF dropoff - see simulation. This effect may be a more likely explanation for "I changed the FET and it sounds different" than mystery un-measurable FET properties.
Etc...
Thanks for the confirmation and the backgrounds of C4! This really helped.
Jan
