Building a Pearl 2

I am pretty sure the 2 boards with fets have always been $200. Boards are Bentley quality! It is an expensive project but well worth it.

Yeah they sure look nice in other owners builds! When i paid for mine a few weeks ago (placed my order) there was an option to either buy a single board with JFETs at 100, or without for 50. today they only have one option for 200$ and it says with 6 FETs.. which sounds wrong...

Wondering if i should-a just bought the boards only. cant imagine 12 matched JFETS are worth more than 6-12 dollars... though the matching process probably requires a lot of excess JFETs before matched.

what other JFETs have been used in this project with good results... i'm not sure what their roles are in the circuit, but im assuming its very instrumental in the sound of the final product
 
Yeah they sure look nice in other owners builds! When i paid for mine a few weeks ago (placed my order) there was an option to either buy a single board with JFETs at 100, or without for 50. today they only have one option for 200$ and it says with 6 FETs.. which sounds wrong...

Wondering if i should-a just bought the boards only. cant imagine 12 matched JFETS are worth more than 6-12 dollars... though the matching process probably requires a lot of excess JFETs before matched.

what other JFETs have been used in this project with good results... i'm not sure what their roles are in the circuit, but im assuming its very instrumental in the sound of the final product

I would say 12 matched fets 2sk170 are worth at least $50. It looks like a new web site with a few mistakes - pretty sure it should be 2 PCB boards for $200 with 12 fets which is consistent with the old web site pricing from a few weeks ago when I was on there.Call them up on the phone to discuss - I talked with nice lady there once - picked up right away when I called to check on shipping.
 
Anyone ever house each channel of the Pearl in small separate enclosures? advantages? drawbacks?

Thanks AV werk. I ordered a 24volt model. Fingers crossed its not over voltage like yours was. If it is should I add regulator chips (redundant) to ensure stable exiting voltage, or use series resistors? Is there another way to shed voltage that might be useful? like a light or screen on something? LEDs?

Dont know if the same is true for R-cores but toroids often have higher voltages when unloaded. My toroid is 2x24V, which should be about 34V rectified, but it's more like 39V. As soon as you draw a few mA from it you can see the voltage dropping. With the power consumption of the Pearl I guess the voltage will be close to 34V again, and thus no problem for the 7824 and 7924. The only get lukewarm in my build.
 
Feel free to evaluate my thinking here.. and interject where you see fit as I am learning.

My goal is to get a very stable, ultra quiet, pure DC supply to ensure the pearl has it as good as It can from a basic PS. I see the psu as a critical part of a phono-amp and not to be overlooked. My brainstorming and reading has lead to a slightly more complicated, but still very basic power supply that is not too dissimilar to what others are using.. but with a few added tidbits.

P-core to offer split bobbin isolation of the primaries and secondaries.
Rectified ground, and a perhaps a small 0.1uF cap between chassis and ground (open to suggestions).
There will be a suppressor on the DPDT power switch, and an integrated fuse.

my plan is to use 2 Reservoir Caps instead of 1. snubbing will be placed immediately after the schottky bridges. I'll be using 2 x 6800uf Elna LAO's paralleled to give a little more than 10000uF. Between them, 4 low value 0.5Ohm resistors will be added in a parallel configuration (to give 0.125ohm) to form a sort-of low resistance CRC filter. Im also using a bleeder for safety and LED's as an indicator for power for each rail. These, I'm hoping, will provide a small constant load on the transformer and keep the current forward through the bridges. Im hoping the LED's will burn until the caps are more or less drained. With the P-core and low load of the pearl I'm doubting heat will be an issue anyways.

Apparently with the diodes i've selected snubbing isn't necessary as they are extremely fast and with very soft recovery, but im not under the impression including them hurts so i will since i have 0.22uf snubbing caps already. (The data-sheet suggests the diodes are very quiet and that snubbing is not necessary... I figure, in case they are just bragging, why not keep the snubbers in.)

Im also planning an additional high frequency filtering step to make for the flattest DC i can. This had me consider inductors. After some thought, Ive decided against inductors to avoid potential issues with grounding and voltage increases... plus, without knowing the -actual- minimum load i cant determine what the optimal inductance values would be. I postulated a guess based on Nelson's suggestion in a power amp psu. With a calculator and the assumption that the pearl draws 100mA, i assume it'd need less than 1 or 2 mH. ... so voltage increase might not be too much, but Not wanting to dive into that type of filtering scheme... I favoured CRC again for high frequency filtering.

... so to keep it simple, For the high frequency dc filter Ive chosen some small value KEMET film caps 0.1, 0.01, 0.001, 0.0001uf with 0.05ohm resistors between (0.05 x 3 = 1.5). This is based on the suggestion that smaller valued caps are more adept at smoothing higher frequency noise in the dc supply, and those electrolytic Elnas tend to be weaker at dampening higher frequency dc bourn noise. Again, low value resistors to avoid excess heat in the psu enclosure and to avoid excess voltage loss.
 
Last edited:

6L6

Moderator
Joined 2010
Paid Member
You do realize that the single biggest thing that makes the Pearl so quiet is that fact that you can place the PSU box 3 or 4 feet from the RIAA box... :) (And that it has more PSU capacitance than some power amps...)

I'm all for your ideas and mods to make the PSU completely overkill and fantastically flat. I'm sure that you will find a few good improvements.

My suggestion, said from experience, is to build it in a simple and 'stock' configuration first. Get the RIAA up and running, happy and burnt-in. In about 1/2 of the builds, people are having some sort of problem, but they get sorted out and things are then good. So get it stable and grounded properly first.

Then mess with the PSU. That way you will know if there was any improvement.
 
I think there-in lays the best piece of advice 6L6...

"build it in a simple and 'stock' configuration first [...] Then mess with the PSU. That way you will know if there was any improvement."

I'm willing to bet you know exactly how i feel, so no peace of advice is lost on me. The more I read about each given component or PSU design in general the more i'm compelled to explore ideas. I like coming up with an idea/design of my own. ..however, if it doesn't work then i wont know if its because of shoddy assembly or the idea itself. i sketched up the filter idea for a separate board so i can just plug it in line to check it out.

regarding the caps and LEDs. .... they're polar of course..

Pos
V+
V0

Neg
V0
V-

... with the negative rail, do i wire them + side up to the V0, oriented just like the positive rail or do i mirror the positive rail (flip em around)? don't want to put these puppies in upside down.

Ac Pulse can go in either way right?
 
So I'm pleased to say I've completed the power supply for the pearl, and a bulb tester (as per directions on decible-dungeon). Im very proud of my first diy so far!
I think i might be having some issues though.. Im hoping its all benign troubles. Can anyone chime in with some Help ... Ive blown both of my fuses.. And will need to buy more.. But before i start blowing more fuses and money... Could i borrow your expert minds to help me sort it. I wonder, Could tripping be due to power up surging as the caps fill up?? Im using 2 x 6800uf LAO series caps per rail with small wire-wound resistors for a crc layout. I have an led on the positive rail to indicate power. There is a bleeding resistor across the second cap of each rail. No snubbing caps installed.

First things first.. So. Im blowing fuses basically right away. The first one lasted about 20 seconds.. Just long enough to see that the voltage on the dc was around 14 or so.. With a 40 watt bulb in the tester. The bulb didn't dim completely, but did run quite dim. It never stopped burning untill the fuse blew. As soon as i hit the by pass to see if the voltage would climb to the target 36v the bulb went out and sure enough it was the fuse - it couldnt handle full power.. I didn't manage to test the ac voltage after the transformer.

The second fuse blew before the bulb would even fire up. So the bulb never lit uo. They were 500mA 250v not sure if they were slow or fast types. Alll the diodes and the measure the same in the diode test.1.8 for the led and 2.4 for the schottkeys if i recall. There is continuity between + and - rail ,... Im guessing thats because of the bleeders....? Ground is on the chassis side of loop breaker.

Transformer is shielded and centre tapped.

+Gnd is wired to Crc bleeder and led was wired point to point on a plain prototylpe board. Diodes are on Peter Daniels board. No snubber- caps yet.

Thoughts? Why the fuses dlowing
 
The transformer is an R-Core and is wired as follows:
1) Two red wires on "110v" primary (Wall voltage is 118v.) wired to switch with xtype capacitor across the live side.
2) One green/yellow wire that the diagram shows to come from between the windings and is labelled with a ground symbol (i understood as a shield) so i wired that to the chassis, bypassing the loop breaker. The transformer is mounted to the case on rubber bushings. There is continuity between the metal base of the xformer and the IEC ground pin.
3) Two white and one black on secondary side. The secondaries are wired to a peterdaniels gain-clone diode board like this:

+ White.
G+ Black *
G- Black(jumped*)
- White.

*Since there was only one black i patched the two zero lines together w a short wire jumper. I found it difficult to find info to confirm that this is correct as most people have dual secondaries.

The loop-breaker is a 35amp rectifier bridge with ~ and ~ shorted, + and - shorted, and a y2 type capacitor and 10ohm resistor parallel across ~ and - (as per diagram on decimal dungeon and article avoiding ground loops)
 
Last edited: