"Sehr gut"
very good, now the first stage finally works as a cascode - and no longer as a wire.
The good thing is that the P3 still plays very well. When you have tweaked something and it works you are in a positive mind and often you think that it sounds a little bit better than before. Same here.......slightly warmer so more neutral now than before if I should say something about the sound before and now. But always very difficult to remember how something sounded. The most important is that I (still) enjoy the sound.
To something very different........the logo of Led Zeppelin as pictured below (I just played the record + a couple other). It reminds of the the ZEN MOD LABS logo. Is that just a coincidence?
There is a font called Kashmir that looks a bit like it.......wonder if i can switch to Kashmir font when I write in this forum.
To something very different........the logo of Led Zeppelin as pictured below (I just played the record + a couple other). It reminds of the the ZEN MOD LABS logo. Is that just a coincidence?
There is a font called Kashmir that looks a bit like it.......wonder if i can switch to Kashmir font when I write in this forum.
Attachments
@Wellerman -- try a couple of things
if the P3 isn't enclosed in a shielded chassis the bipolar opamps (LM4562, LME49720 etc) may pick up noise from external sources -- anything from a nearby switch mode power supply, AM radio station, LED lamp etc.
I doubt that it's the opamps. I have not seen the problem with LM4562, OPA2604, LT1113, OPA2134 or NE5532. The input-voltage noise on my P3 is slightly less than shown in the schematic as I used 4x 2sk170BL.
if the P3 isn't enclosed in a shielded chassis the bipolar opamps (LM4562, LME49720 etc) may pick up noise from external sources -- anything from a nearby switch mode power supply, AM radio station, LED lamp etc.
I doubt that it's the opamps. I have not seen the problem with LM4562, OPA2604, LT1113, OPA2134 or NE5532. The input-voltage noise on my P3 is slightly less than shown in the schematic as I used 4x 2sk170BL.
Hello,
On one hand theory is nice, however if you make a cognitive error you end up with a incorrect assumption.
On the other hand you have practical hands on measurement.
Connect your analyzer output to the P3 input. Connect the P3 output to the analyzer input and watch the "Gain Meter" (calculated by the analyzer) vary as the value of R1 is ramped up or down. Yes the total gain does ramp up and down with the gain of the input stage.
Thanks DT
Last edited:
A small amount, sure, but not enough to care about. There is after all only around 17dB excess gain in this design on the MC setting. So front end gain changes of a dB or two are reduced by a factor of ~7 by the feedback. Is there really a concern about a fraction of a dB gain change?
My P3 is still under construction, and I'll be mucking with it as I go. But nothing to measure just yet. In the meantime if you have some measurements to share, please do. I'm happy to correct any erroneous statements.
My P3 is still under construction, and I'll be mucking with it as I go. But nothing to measure just yet. In the meantime if you have some measurements to share, please do. I'm happy to correct any erroneous statements.
Last edited:
I've been really enjoying my P3 since completing it in December. On completion, I performed gain, FR, and distortion tests with my QA403, and those measurements were very close between the two boards.
It has been functioning perfectly until a day ago when I started noticing a drop in gain in one of the channels. I've measured rail voltages (A-OK), and at a few other nodes, including the differential BJTs (close), but voltages on the front end FETs are now lower on one board than the other (both matched pretty closely when built). Voltages across each 2KS209 Rs on each board are now different.
When I have time to get back to this tomorrow, I'll be verifying voltages at other nodes are proper, including the op amps, then focusing on the input FETs. Might need to get out the hot air pencil...
It has been functioning perfectly until a day ago when I started noticing a drop in gain in one of the channels. I've measured rail voltages (A-OK), and at a few other nodes, including the differential BJTs (close), but voltages on the front end FETs are now lower on one board than the other (both matched pretty closely when built). Voltages across each 2KS209 Rs on each board are now different.
When I have time to get back to this tomorrow, I'll be verifying voltages at other nodes are proper, including the op amps, then focusing on the input FETs. Might need to get out the hot air pencil...
I gotta say it. Those of you fussing over cascode bias and theoretical noise floors kinda baffle me. Since I've put my Pearl 3 in system I have replaced a stylus head, bought an aftermarket platter, put my turntable on isolators, settled on a single preamp (a B1 buffer only- everything else is 100% unnecessary gain-wise*), revised and gone back and forth several times on the crossovers in my speakers, and listened to all 4 of my favorite amps and god knows how many records and genres in system without reaching any firm conclusion as to how my system "should" sound. I haven't even rolled past my second op-amp. This phono stage, which is the most revealing pin-point thing I could imagine (a friend with golden ears called it a "flamethrower"), has shown such a spotlight as to call everything else in system into question. Are some of you just not playing records rn?
(*this is a Pass B1 buffer, turned to 3 O'clock, sitting between a Pearl 3 and an Aleph 20. It's too loud any higher. Throwing away buckets of gain.)
(*this is a Pass B1 buffer, turned to 3 O'clock, sitting between a Pearl 3 and an Aleph 20. It's too loud any higher. Throwing away buckets of gain.)
Hello,
I generally take the Pete Millett approach that may not include listening.
Of the equipment that I listen to, only the speakers are made by me. Only after a lot of poking, prodding and measuring.
Building and measuring is one kind of fun. Enjoying the music is something else.
The photo of your B1 buffer shows the volume knob at 9 O'clock, only turned up a little from the minimum position. Too many buckets of gain.
Before I plug in the P3 and listen to the music I need to sort out how to get the total gain down to 40ish dB from the as measured 50+dB's that it is now. The P3 is installed in a 8 X 8 X 4 inch 16ga steel J-box.
The stuff that I listen to on my bench is a upgraded sound card in my computer, a used NAD THX 216 power amplifier repaired by me (99 SINAD at 5Watts output) and for speakers 6-1/2 in Purifi mid-bass + beryllium tweeters built and measured by me. With a 12in Peerless sub.
The main system in my space rotates, now it is a Benchmark power amp, Benchmark pre/headphone amp, Topping DAC / Bluetooth , old used Kenwood radio tuner and a Cambridge CD player. The current phono pre is a used Parasound thing. The speakers are some ancient JBL monitors. The last speakers were stacks of active tri-amped, 15in woofers, 10in mids and horn tweeters. love the displacement but unfinished MDF does not look so good.
The finished P3 looks so nice that it may find a space on my shelf.
Thanks DT
I do like to listen to the Grand Ole Opry on the Harman XM radio in my car or truck.
It is part of the fun. And in case of the cascode: with high Idss parts for the jfets it is part of making the P3 work as intended. For musical enjoyment I have the Pearl Two. I rebuilt or tweaked that one a few times during 10 years before its final incarnation that I am enjoying now.
For illustrative purposes only, a little bit over the thumb.
R1||R10 = R_drain; R1_max(GR) = 750 Ohm & R1_min(BL) = 470 Ohm; R10 = 22kOhm
0.0436 A/V * 725.3 V/A corresponds to 30dB (+33dB) -> OLG approx. 63dB
0.0436 A/V * 460.2 V/A corresponds to 26.05dB (+33dB) -> OLG approx. 59.05dB
Now we close the negative feedback loop,
(1412.5/1413.5) * R9/R8 -> 47dB
(896.4/897.4) * R9/R8 -> 47dB.
The first stage of the Pearl 3 is really very interesting and the insane loss of 4dB in the open loop can, if you dare, be compensated by an R10 = 13k7.
It simply doesn't hurt to think beforehand and only plug in the analyzer later.
Thats the miracle of negative feedback, what about the distortions?
Let's try some theory right away:
R1||R10 = R_drain; R1_max(GR) = 750 Ohm & R1_min(BL) = 470 Ohm; R10 = 22kOhm
0.0436 A/V * 725.3 V/A corresponds to 30dB (+33dB) -> OLG approx. 63dB
0.0436 A/V * 460.2 V/A corresponds to 26.05dB (+33dB) -> OLG approx. 59.05dB
Now we close the negative feedback loop,
(1412.5/1413.5) * R9/R8 -> 47dB
(896.4/897.4) * R9/R8 -> 47dB.
The first stage of the Pearl 3 is really very interesting and the insane loss of 4dB in the open loop can, if you dare, be compensated by an R10 = 13k7.
It simply doesn't hurt to think beforehand and only plug in the analyzer later.
Thats the miracle of negative feedback, what about the distortions?
Last edited:
Sorry,
in order to make a direct comparison (possible), I still have to use the same method. In the first post the slope was 43.6mS and in the second post 11.85mS.
Now we can look directly at the influence of both the resistance value R1 and the influence of the jFets used on the resulting gain factor in a counter-coupled manner. 47dB (rounded) at the top to 46.83dB at the bottom and again 46.81dB at the bottom. No significant difference. In fact, we only observe 42.8dB(750Ohm) to 41.5dB(470Ohm). +1.3dB on the plus side for the larger R1. The difference to the open loop is now |8.7dB| with 750Ohms and |6.77dB| with 470Ohms.
If Wayne allows us to, we could reduce R10 and tentatively increase R8 with a clear conscience.
😉
in order to make a direct comparison (possible), I still have to use the same method. In the first post the slope was 43.6mS and in the second post 11.85mS.
Now we can look directly at the influence of both the resistance value R1 and the influence of the jFets used on the resulting gain factor in a counter-coupled manner. 47dB (rounded) at the top to 46.83dB at the bottom and again 46.81dB at the bottom. No significant difference. In fact, we only observe 42.8dB(750Ohm) to 41.5dB(470Ohm). +1.3dB on the plus side for the larger R1. The difference to the open loop is now |8.7dB| with 750Ohms and |6.77dB| with 470Ohms.
If Wayne allows us to, we could reduce R10 and tentatively increase R8 with a clear conscience.
😉
OK, problem solved. No problems with voltages or components. Main issue was Q1 needed to be re-flowed. However, after doing so, I found another issue. The pair of RCA cables connecting my 1k sine wave generator to the Pearl 3 were also causing a small signal drop on one channel. Into the trash bin with those!! Re-testing with good Pomona BNC cables showed almost perfect channel balance.
More than 24 hours with this out of the system - what a drag! 😢
More than 24 hours with this out of the system - what a drag! 😢
A different strategy with regards to the cascode is to make R3 4.75K. The voltage is lower on the drains but at the AC levels it isn't much of an issue.
I don't see much if any measured difference with any of these resistor or transistor changes.
I just built mine per the schematic and have been listening to it since September.
I don't see much if any measured difference with any of these resistor or transistor changes.
I just built mine per the schematic and have been listening to it since September.
About time for an update! Kids, works, etc... So I was able to eliminate all chassis hum, which is good. I attached the shield ground to chassis ground, removed some more paint/anodizing to ensure the chassis was attached, upgraded to a new tonearm cable, and grounded the bottom of the turntable to the Pearl 3 in addition to the normal junction box to Pearl 3 ground. However, I isolated some hum (at very high volumes) to the VPI tonearm junction box. When I touch the junction box, boom, hum gone. If I touch the tonearm, hum gets louder. So I'm pretty sure that's my issue. Now to figure out how to fix it, as it seems like a common issue with some VPI tables. If anyone has solved this issue please let me know. I plan to email VPI as well. Thanks!
Just a curious question regarding the cascode and saturated Q5. It was mentioned that if Q5 is saturated then you could insert a wire instead.
Has anybody tried that or simulated it?
I won't try it but if a wire is the same as Q5 saturated then R2/R3 could also be omitted?
What I would like is to read about the theory behind an amplifier with cascode.
Has anybody tried that or simulated it?
I won't try it but if a wire is the same as Q5 saturated then R2/R3 could also be omitted?
What I would like is to read about the theory behind an amplifier with cascode.
Couldn't agree more. Many prefer to paint by the numbers, and there's nothing wrong with that. You'll get a great piece of functional art, with a high probability of success.
I like to color outside the lines. This is DIY after all - I think some experimentation should not just be tolerated, it should be encouraged! There's not much at risk with a little preamp circuit like this. And the music will be all the sweeter when it's time to listen.
You can raise the value of R18 to reduce the gain without altering the RIAA EQ. Swapping from 220 Ω to 680, 690 or 706 Ω will knock the gain down by ~10dB. I'd also reduce the value of C8 in proportion to maintain the LF pole. A 68 µF bipolar would be a good choice here.
I again learned something from this last post. I made R18 270R with C8 = 100uF to get my desired LF roll of, but did'nt realise it would also alter the gain. But I guess the difference between 220 and 270 is not that big?