DIY-Central.com - Go Create something! - PCB Design, Electronics

Did the noise floor at max gain drop at all?

By -1dB! But it's probably awash. It could have been better.

Using the external gain switch with about almost 6" of wires (3" going to, 3" coming back) made RF noise worse. See that clump of white wires there going to the switch?

But "grounding" the case made the unit quieter. In fact, it's a different kind of quiet hissing sound. And again, you only hear this hiss sound at max volume and at +66 and +72dB gain. Anything below that gain and it's super quiet/silent.

Maybe that's why a certain commercial preamp's max gain level setting is only 59dB! I won't mention names. :)

My recommendation: On the final kit, don't save a few bucks to buy a $6 plastic rotary switch instead of the $16 Bourns PCB-mounted, metal selector switch originally spec'd. The wiring job is tedious.

See that big clump of white wires in the photo? Touch it and you'd pick up all sorts of AM radio. (+72dB gain = 4000x amplification!)

But I do know some people are interested in using hookup wires instead of the onboard selector switch so I had to try it. It could work, it will just take more time assembling it, and there could be more potential noise/RF problems. Neat wiring and soldering job is a must if going this route.

Right now, I measured noise floor at -63dB and -64dB @ +72dB gain.
And -70 to -69dB at +66dB gain.
-76 to -75dB at +60dB gain.

photos at +72dB gain.

FINISHED!!!!!!!!!

Worked on this project this Friday, Saturday and Sunday.... whew!

Since I used an external 12-pos switch, I basically had to solder 13 wires, so that equals 26 solder points per channel, or 52 solder points for 2 channels!!! I wouldn't recommend this approach. The wires are also prone to RFI pickup. I made the wires as short as I can but still, it's prone to pickup interference.

I would recommend you guys use the onboard PCB selector switch instead.

To fix the RF problem at 66dB and 72dB Gain, I "discovered" a new (new for me at least) of fixing ground loop/rf problems.

I wired the ground of the 1st channel to the XLR output gnd of the 2nd channel and vice versa. That lowered RF problems, but there was still some slight noise but I noticed that if I parallel the 2 gnd wires and stick them flat against the case, it fixed the RF problem. So that's what I did. You can see the "black tape" holding the 2 gnd wires flat against the case.

I don't know what to call this method... It is definitely NOT a star ground. It's more like an X ground.

Now, the only thing you hear is normal "hiss/white noise" at 100% volume @ 72dB gain! There also seem to be a spike at 15.7Khz. But at normal listening volume (unity gain on mixer), the hiss/white noise is only slight even at 72dB.

Some photos!!!

From this....

To this....

Some close-up photos.

 

 

 

 

 

 

 

Rev3.00 PCB will be coming in middle of this week from the PCB manufacturer, and then I'll test those samples by building prototype #003 and #004.

April 20, 2007

I'm racking prototype #001 and #002. Getting ready to use it for sample recordings, so I need to be able to take it anywhere easily. I'd also like to see if racking it in a metal case will improve noise floor further... hope so!!!

Also going to test how the 7818/7918 PSU type regulators stack up vs. 317/337 type regulators. I want to see if 317/337 are really better than plain stock 7818/7918, or if it's just a myth.

This is still the PCB without the onboard pots and selectors. So I guess, we'll also see noise performance when using external pots and switches with hook-up wires.

Here are some photos.... sorry for the harsh shadows, I just used a single flash on the right side and didn't use any reflector on the left.

If you use on-board pots and switches, you can easily fit 4 preamp channels on a 1u rack... with plenty of room to spare in the back.

http://www.fivefish.net/diy/sc1/images/protocase2.jpg

The front view of the prototype rack. (On the background, you'll see my new "electronics assembly" factory. I'm getting ready to package some PSU kits to sell.)

Closeup of prototype #001.

The switch you see in the picture is just a jury-rigged DPDT switch using component leads soldered to the switch and PCB.

Closeup of prototype #002.

Knobs and switches and pots on front panel.

Back view showing IEC AC line filter connector, fuse, and Neutrik XLR male and female jacks.

Getting ready to build a PSU for this. I already have an 18-0-18 designed PSU, but I think I can still improve upon it. I'll try my Rev1 PSU design, and compare it with a 317/337 type psu. See which one is better.

The PSU kit will be sold separately from the SC-1 kit.

Current consumption for (2) channels.

+/- 26.4 mA (2 channels, with phantom power LED OFF)

If phantom power is ON, the LEDs consume a few milliamps.

if using a diffuse RED LED, it's consumption is 3.1mA x 2 = 6.2mA
total = 32.6mA

if using a clear RED LED, it's consumption is 4.9mA x 2 = 9.8mA
total = 36.2mA

Add another LED for the power on switch indicator... 4.9mA

And estimate about 10mA to 14mA for each phantom powered mic attached = 28mA

GRAND TOTAL ESTIMATE: 69.1 mAmps for 2 channels

To be on the safe side (with 100% safety factor), let's round it up to 140mA.

So the VA needed will be 18V x .140 Amp = 2.52VA

Looking at available toroidal transformers, we can buy a 3.2VA transformer (18V @ .178 Amps) - Good enough!

But if you'll be adding additional modules in the future, maybe the 7VA transformer (18V @ .388 amps) is a much better buy considering it's only $1 more expensive compared to the 3.2VA. - Recommended!

The SC-1 has one feature that I have not talked about much. It's got provisions for an INSERT.

Just like the Inserts in your mixer, the concept is I want to be able to add future signal processing to the SC-1 easily.

I envision in the future creating add-on modules to the SC-1. For example: Passive EQ, Active EQ, optical compressor, VCA compressor, etc... so all you'll need to do is solder 3 wires, remove the jumper JP2 and the add-on module is now part of the signal path.

If you're a purist, and you don't want anything else changing the sound... just don't add any modules.

If you want a preamp with an EQ, you can have it.
If you want a preamp with built-in compression, you can have it.
If you want THE WORKS, preamp with built-in compression, EQ, etc...

Just add the modules you want. You can see the 3 solder pads in the PCB in the photo below, near the lower left corner.

As far as the insert modules go, would it be possible to wire this as a 1/4" insert jack?

I looked into the datasheets, and it could work.

The SC-1 will have a 5K input impedance for the insert return. And the send can have a max +/- 13V swing (on a 15-0-15 supply) output voltageon a 2K load.

Note: the SC-1 will be powered by 18-0-18 so the output voltage swing could be higher.

(The Mackie 1402VLZ has 2.5k input impedance for the insert return. )

I'll try it out one of these days.

The saga continues... Time to get picky! I want to stamp out every possible problem that could happen.

Hooked up the preamp via 20ft xlr to trs cable. From the test preamp it went into a patch bay, then a 3ft cord, then to MOTU inputs.

Used Digital Performer to record the tracks and measure the "noise" DP registered in it's VU meters.

Gain position - Gain dB - VU meter reading
============================
1 - 0dB -90dB
2 - 6dB -90dB
3 - 12dB -90dB
4 - 18dB -90dB
5 - 24dB -90dB
6 - 30dB -90dB
7 - 36dB -84dB
8 - 42dB -74dB
9 - 48dB -66dB
10- 54dB -59dB
11- 60dB -52dB
12- 66dB -44dB

As you can see, noise started to increase at the 8th gain position -74dB reading for a 42dB of gain.

During my testing the other day, I did notice the oscillation on the 5th, 6th and 7th position which was reduced/eliminated by the bypass capacitors.

But the 0.1uf bypass caps were ineffective for gain position 8th and higher. ... and it shows in the readings above.

-74dB is still not too noticeable, but -59, -52 and -44dB are very high.

So I need to find a solution for this and get rid of this problem. I'm thinking a different value of bypass cap in parallel with the 0.1uf may work.

So for now, Rev3.00 PCB is on hold for manufacturing.

UPDATE
Incorporated some fixes. Added more bypass capacitors to the volume potentiometer. I think considering the volume pot is connected up via hookup wires instead of onboard the PCB is a contributing factor to the noise problem.

Picked an AM station when the preamp is hooked up directly to the RP8 monitor. Didn't happen when the pre was hooked up to the MOTU. Added another bypass cap to the volume knob and that solved the problem. Bye bye country radio!

Gain position - Gain dB - VU meter reading - New reading
============================
1 - 0dB -90dB -90dB
2 - 6dB -90dB -90dB
3 - 12dB -90dB -90dB
4 - 18dB -90dB -90dB
5 - 24dB -90dB -90dB
6 - 30dB -90dB -86dB
7 - 36dB -84dB -84dB
8 - 42dB -74dB -78dB
9 - 48dB -66dB -78dB
10- 54dB -59dB -65dB
11- 60dB -52dB -60dB
12- 66dB -44dB -55dB

DP's VU meter readings are better... especially at the 11th and 12th position. Oscillation at the higher gain settings fixed too.

Just for some layman terms...

60dB is 1000x amplification of the original signal.
66dB is 2000x amplification of the original signal.

BTW, without the preamp connected the best my system can do is -90dB. So the above -90dB readings probably should be lower.

I still want to be able to push noise lower for gain settings 54, 60 and 66dB maybe down to the -78dB mark.

The thing is since I'm using about 2" jumper to simulate my selector switch, most of the noise is coming in via this jumper. I can tell because depending on how the jumper wire is positioned, or if I'm touching it, the amount of noise varies. If I had a real selector switch soldered in, I know the readings will be even lower than above.

Another possible area where noise is coming in is via the hookup wires that connects to the external potentiometer. This piece of wire then goes directly to the balanced driver stage.

On Rev3.00 PCB, the gain selector switch and volume potentiometer will be onboard the PCB so we're talking about very minimum distances, no hookup wires and very close to the ground plane of the PCB. I don't think we'll have this problem on the Rev3 design.

So maybe I might go ahead and push to manufacturing...

more update
I decided to plugin again the SM57, crank the gain selector all the way to 66dB, and lowered the volume knob since the SM57 doesn't really need that much of a gain... hooked the preamp directly to the RP8 monitor and fired it up.

Better than yesterday and earlier today! It is quiet. The volume knob pot is like 1/5th up and it is already loud, of course considering it's at 66dB gain (amplifying 2000x) and no RF noise, no hum noise... YES, at 66dB setting!

Noise measurements:

I'm a little bummed that at the 60dB and 66dB gain settings, the SC-1 noise floor isn't lower than I expected. (@60dB, -52dB improved to -60dB; @66dB, -44dB improved to -55dB)

So I looked around the net at other equipment specs...

There's a lot of mumbo jumbo out there... Some measuring their noise with the inputs shorted to ground (which personally I don't agree) and some not specifying what's the volume level at (i.e. you can set gain to 66dB, but if you leave the volume only at 50% can get a low noise floor reading... deceiving don't you think?)

I found a Rane note (http://www.rane.com/note145.html) that says GAIN should be set to maximum (I did that) and the input TERMINATED with the expected source impedance. (Oh oh.... I did NOT do that.)

So, I goofed up. When I did the earlier noise measurements, I left the inputs hanging.

It should have been terminated with a 150-ohm resistor to simulate a mic connected to it. So I got a spare XLR jack, soldered a 150ohm resistor on pins 2 and 3, plugged it into the mic preamp and measured performance again.

WOW! Big difference! Much better.

(This also explains why the other day when I hooked up an SM57, it seems quiet and great, but when I checked the noise floor in DP (with inputs hanging), the readings were high.)

So here are the NEW and improved numbers!

Gain dB = Noise floor (Rs=150 ohms, Max Gain settings, volume at max)
================================
0dB = -90dB*
6 dB = -90dB*
12 dB = -90dB*
18 dB = -90dB*
24 dB = -90dB*
30 dB = -90dB*
36 dB = -88dB
42 dB = -85dB
48 dB = -80dB
54 dB = -74dB
60 dB = -69dB
66 dB = -64dB

* Note: The best my setup can do is -90dB, so I cannot measure anything below this.

** These are not EIN numbers.

And yes, the above values are at Max Gain, Volume knob all the way to the right - settings. This is real-world results, not theoretical calculated noise figure. This is what you'll see on your VU meters in your DAW software.

So this is the current score... we're at -64dB noise floor at 66dB gain.... much better! Now, the challenge is to reduce this further. But I'm just nit-picking here.

Noise Obsession

I've been pondering this noise issue for a couple of days now.

...... so last night, I'm trying to review my design and reviewing the data sheets....

I *KNOW* I have the resistor values for the mic pre stage I have were computed for 0 to 66dB gain. Then it goes to a servo, then to an unbalanced to balanced line driver.... which then goes to the MOTU inputs.

So I'm thinking... maybe it's the line driver stage giving me the noise problem??? The 1510 chip.

So I looked at the datasheet and something caught my eye.... the line driver stage ADDS ANOTHER 6dB GAIN!!!! DUHHHHH!!!! Of course! Going from unbalanced to balanced, gives you an additional +6dB gain!

So really, my preamp does not go from 0-66dB in 12 steps, BUT RATHER it goes from +6 to +72dB gain!!!! (i.e. 66dB mic pre gain + 6dB gain of balanced driver = 72dB total gain!)

So the correct noise figures compared to dB gain are:

Noise Floor at max gain settings, max volume, 150 ohm terminated inputs

6dB = -90dB* 12 dB = -90dB* 18 dB = -90dB* 24 dB = -90dB* 30 dB = -90dB* 36 dB = -90dB* 42 dB = -88dB 48 dB = -85dB 54 dB = -80dB 60 dB = -74dB 66 dB = -69dB 72 dB = -64dB

So there you go!

The FINAL SCORE!!!....

-69dB @ 66dB Gain with Rs=150 ohm terminated inputs, volume at max 100%. Which is about comparable to the RNP's noise floor and other non-transformer based IC mic pre design.

And for the icing on the cake, mine can go up to 72dB of gain, not just 66dB!!! -64dB @ +72dB gain (which is 4000x amplification!)

Not bad!

Found an oscilloscope plugin...
http://mdsp.smartelectronix.com/ffttools/2005/03/freakoscope-08-public-beta.php

continuing with this noise obsession... using the above plugin.

Since a picture is worth a thousand words.

Update:

Well, I spent some time working with the prototype.

Noticed some slight oscillation at the 5th, 6th, and 7th gain settings of the selector switch. So finally fixed that with additional bypass capacitors on the supply rails. We're talking about less than 0.2Volts oscillation on top of a 30Vpp output. So it's insignificant. But still, it's gone now. :)

As I said, this is Rev1.00 PCB I'm working on and it looks like on my Rev3.00 PCB, I've already incorporated those caps in the design.

Then connected the mic pre circuit output to the balanced line driver. And what do you know... the less than 0.2Volt oscillation fixed itself. The addtl caps weren't needed.

I can't wait any longer, and it's time to hear this with my own ears. So I connected my RP8 monitor to the preamp output jacks. And run the signal gen from 20Hz to 20Khz. (I didn't run it higher than that since I might burn out my tweeter!) The RP8 cannot reproduce 20Hz, and I only start to hear sound around 30+Hz... makes sense since the RP8 freq response is stated at 45Hz to 20Khz.

It passes audio, sounds great! Gain selector switch working. Even at max setting of 66dB!!!

Connected a dynamic mic. Very clean even at high gain settings. I sometimes thought the unit was off because I can't hear any residual noise when I have my ears near the RP8. Then I tap the mic and yup... it's on and working.

Next, passed some music material. Very clean, no hum, no noise .

Played "Misssing (Remix version)" by EBTG. Woohoo... the bass is so deep and clean and distinct. Not muddy. Played some synth heavy music. The highs and mids are very clear too. And quiet! No hum or noise.

I think I'll probably build a 2nd prototype of the Rev1. PCB so I can listen to it on stereo! :)

I still need to test this using a condenser mic. But from all signs, I think it will work just fine and just be as good.

More update: Condenser Mic Testing/SM57 Mic Testing:

Tried testing with some condenser mics today. Niiice! I just used +18V for the 48V phantom power since my bench psu doesn't have 48V. You don't necessarily need to have +48V for your condenser mics to work. The mics all worked just fine even at +18V, no problem!

As for the sound... it really sounds very full and very nice.

And I was blown away when I plugged in an SM57 dynamic. It *really* sounds great. Not boomy or thin, not harsh, but you can hear everything clearly.

I don't have any high-end condenser mics, but I can tell you... even a cheap MXL 990 shines on this thing. Very clear. Lots of response.

And yes, also tried a $20 Nady dynamic mic. It's not as great as the SM57, but it's still clean and quiet. A little lacking on the high end compared to the SM57.

Rev3.00 PCB Prototype

So what's next?

Rev3.00 PCB Design

I'm going to send PCB Rev3.00 (shown above) for a prototype run, build a couple of units using the Rev3.00 PCB, check the final parts list and bill of materials... and if those units work great (which I'm sure), I'm going to send the PCB out for mass manufacturing... mass as in maybe 100-250 PCB units. I'm just a small-time potato running this operation in the garage.

What's new in Rev3.00?

1. Added J.W Miller inductors at the output for RFI protection
2. Added coupling caps for balanced line driver in the output stage
3. Added jumpers for easy setup
4. Added coupling capacitor in case you don't want a Servo design... or you want a Servo + AC coupling setup. Added jumper for those who don't want AC coupling.
5. Change value or Resistor for LED phantom power indicator. Made the LED brighter. (it was a bit dim on Rev1.00)

Ordering Info

I'm shooting for mid May to end of May when you can start ordering them.

So start saving!!!! For sure, it will be less than $100 per channel (for the kit version, with all the PCB and parts included. Just supply your own soldering iron.)

Discounts will be given for those purchasing 2 channels, or preamp and psu combo. Bigger discounts for those wanting to purchase 8 channels.

The SC-1 preamp worked and it sounds great, very quiet, very clear, with a great freq response.

A lot has happened to the SC-1 mic preamp that I'm developing... that I didn't have time to update this site. Sorry about that.

Here's some new materials/information regarding the SC-1 mic preamp kit.

Basically, I'm getting close to when I can start selling the kits. I made 2 channels of prototype using the Revision 1 PCB design and I can tell you, I'm more than pleased with how it turned out. 

I also updated the design, added more enhancements and some small fixes (mainly, additional bypass capacitors, RF protection) and came up with a Revisi