that 1512/1646 pcb layout - newbie alert! ;)

[LG]

Hi folks,

I've been trying to learn to use eagle cad, running the free version with limited board size. Could some of you kind people answer some questions?
Since the limited board size I've decided to do a double side layout. Im running power lines on one side and the rest on the other.

Is it ok to run a ground plane on both sides on a double sided pcb for filling it up? than i wont be etching away a lot at copper. Are there any reasons for keeping the ground plane on only on one side?
If i'll go on and add a ground plane on both sides, how many vias should i use for connecting them to gather?

Or is this total bullshit? I guess i could route the power lines on the bottom as well with extended use of jumpers.....

The schematic is from the from the data sheets +2 relays for pad + polarity. I intent to change those connectors to one 8 pin molex later and make a pcb layout for the switches and gain pot as well.

Hope someone has the patience to help a newbie

all the best, Leifur

 

[Harpo]

Not a fan of ground planes for non-HF pcbs.
Maybe allow some remarks
Your top/component side layer red pcb traces connect to parts where you couldn't solder on, so this pcb layout would require plated thru holes. Probably not a home etch job.
Looking at caps values you might revisit some caps for their outer dimensions and pin spacings.
I'd rather place the big/missing CG cap on the probably horizontally mounted main pcb instead of on probably vertically mounted gain pot pcb at least for mechanical reasons.
Your C7 (C3 from datasheet) connects to 0V reference voltage instead of shield/XLR-pin1/phantom current return (different symbol and not neccessarily the same). Ideally C5, C6, C7 would be placed where signal enters your box at/close to the XLR-in.
I'd power the relais coils between +/- rails with voltage dropping resistor in between and keep audio 0V reference separate from dirty ground at your decoupling caps, YMMV.

 

[LeeYoo]

Hi.
Is this from your own schematic diagram?
Maybe add buffer caps and 18v transzorbs to the power rails on that board, otherwise the THAT chip will fail the first time you hotplug a mic.
Have a look at the board I have designed.

Google: "Mic Preamp x4" (red circuit boards)

Leo..

 

[LG]

thanks for the replies!

this is just combination of the 1512 and 1646 data sheets suggester schematic + few thing i picked up on the forum..

Here is a diagram as it looks now. @Harpo i'll bring back the CG cap do you have any suggestion what capacity to use? As i recall the data sheet suggest 6800 uF.

Regarding powering relays, I was thinking of using 12vdc HS version since it can opperate on 21.6 max dc and Power
consumption is wery low. Is this wrong? is it better to use voltage droping resistor and use, let's say the 5 voltage rated version of the G5V-2?

now i've made a singel side version of the layout with the use of 5 short jumpers. Still working on making it a bit less crowded since if i get so far that i'll build the preamp i will be home etching i guess.

Leifur

 

[bruce0]

seems like the relay coils are going rail to rail now as Harpo suggests, but they will see 30V without a dropping resistor.
As the xlr is wired with tightly twisted pairs to the board c5, c6, c7 should be tacked onto the xlr so it is near the entrance.

 

[Harpo]

Here is a diagram as it looks now. @Harpo i'll bring back the CG cap do you have any suggestion what capacity to use? As i recall the data sheet suggest 6800 uF.

Assuming the audible range of interest above 20Hz, you might calculate for a -3dB cutoff @2Hz in order to keep phase response near -5° when tested at 20Hz. Cap value for this 2Hz cutoff would be 1/(2*PI()*5R*2Hz)=15,915uF on paper. You might argue, this is worst case condition at max.gain setting (=min.Rg) and you are not going to record infrasonic sources miles away. The 6,800uF value from That datasheet set this -3dB cutoff at 1/(2*PI()*5R*6.8E-3)=4.68Hz. Testing this at 20Hz will give a phase response of -ARCTAN(4.68Hz/20Hz)*180/PI()=-13.17°. In your shoes, I would drop the idea of using a negative logarythmic taper pot and use a switched series string of resistors for a recallable and more predictable gain setting. This way you might not need a Cg cap at all.

Regarding powering relays, I was thinking of using 12vdc HS version since it can opperate on 21.6 max dc and Power
consumption is wery low. Is this wrong? is it better to use voltage droping resistor and use, let's say the 5 voltage rated version of the G5V-2?

With +/-15VDC=30VDC differential supply, using the G5V-2-H1 with 24V coil rating and without dropping resistor will increase rated current demand of 8.33mA only by 2mA to 10.5mA and operate within parts limits. Both of your 12V or 5V H1-types will need a sufficiently rated voltage dropping resistor at a higher current demand, pcb real estate and increased cost.

 

[LeeYoo]

Did you read this paper, page 17: "AES7909_48V_Phantom_Menace_Returns.pdf"
The latest protection sceme suggests 18v TVS diodes and  a minumum of 47uF on the supply rails.
A phantom fault (hotplugging) dumps the stored load of the input caps onto the supply rails, spiking them to e.g +15v and -50v.
The 1510 needs 3300uF, and the 1512 needs 6800uF for the 60db gain settings. Very little DC on there.
I use 3300uF/6.3v computer motherboard caps for my 1510s
Can't find any 6800/6.3v online. Maybe two 3300s in parallel?

I have some spare doublesided 4xPre + supply/PPM board combos soon for $27 shipped.
Newer verion has output protection diodes.
I don't use relays because, in my experience, they will fail after a few years.
Leo..

 

[LeeYoo]

The 2x15v, phantom supply tripler, and 4x PPM 10-LED bargraph driver board.

 

[LG]

As the xlr is wired with tightly twisted pairs to the board c5, c6, c7 should be tacked onto the xlr so it is near the entrance.

Will useing a an shielded cable with pin 1 wired to chassis work in this case?

@ LeeYoo  nice job! wery compact!

I don't use relays because, in my experience, they will fail after a few years.

I realy want to use them, mainly becaus for me this is an exersice and i want to learn how to implement them

Thanks for the note about the Phantom_Menace_Returns. I've added that to the schematics!

Looking at caps values you might revisit some caps for their outer dimensions and pin spacings.

I've been browsing alot trying to find the proper dimensiions. I'm still unsure about the proper minimum voltage rating of the caps. Is there some one who could help with that? I guess the save thing to do would to use around 48V (or maybe 50/60v) rated caps all the way trough, but since there is a space issue smaller size caps would be appreciated. (or I'll drop the stubbornness and buy a Eagle licens )

for the bipolar caps, What do you use? I  was thinking wima or some other poly caps? If I bring in the CG, is that supose to be polariesed or bipolar? I've seen both in various designs.

I've attached a screenshoot from my work in progress, please feel free to comment

@everyone: I really appreciate your help!

all the best, Leifur

 

[LeeYoo]

The 3300/6800uF gain cap has to be polar, because of size/price/availability.
There is little or no DC on that cap on my prototype boards.
Polarized caps should be able to handle up to 1volt reverse voltage without a problem.
You could also measure DC on the cap of the completed amp, and If DC is reversed, turn the cap around.
Input caps have to be 63v, because of leak current (offset problems). I use 47u/63v low ESR caps here.
The 10uF common mode offset caps on the output of the line driver are not in the audio path.
Because they are in the higher impedance feedback lines, they have "virtually no effect on performance" according to the DRV134 datasheet.
Best to buy the caps you are going to use. Then you know the exact physical size for Eagle.
I do not understand why THAT has used 1N4148s in the supply lines.
Not a problem with a running amp (25mA), but will they survive a powerup (charging caps).
I would have used 4004s or 1A schottkys there.
Leo..

 

[bruce0]

re: 1N4148's in in the schematic, it seems that the protection in the schematic presented is based upon the section 3.7 of http://www.thatcorp.com/datashts/AES7909_48V_Phantom_Menace_Returns.pdf where they went to an extra-ordinary amount of detail in managing fault currents from phantom power.

Not sure which diode you are talking about in the supply line.

LG please note, the 1n4004 input clamp diodes - that same article, and others, recommend you use GP diodes (glass passivated) on the input side anyway because they generally have less leakage and thus put less noise/distortion onto pre-gain input.  Not really sure what glass passivated means, but you can find them easy and cheap, and same form factors.

 

[LeeYoo]

Hi bruce0.
The diodes are D7 and D8 in Figure24, page 17.
Their function is explained just above the drawing.
AFAIK, glass passivated just means "preserved" in glass. Like the common 1N4148.
To protect it from environmental degradation.
I use LL4004GP SMD diodes on my boards.

Leifur, Something is wrong with the power rails on your board. D13 is shorted.
Leo.

 

[bruce0]

Leo

I see D7 and D8 and I agree with you, these are typically implemented as protection against reverse polarity hookup and also to deal with cap discharges back to other components on the same psu ... As such they sometimes have lots of different caps hanging across them depending upon the design and how the reservoir caps get populated ( they spec 47uf but 100 might fit or even 220 and if that's all I have I would just say ... More us better it is just stabiliIng the rail .... but I would forget to consider the cap charging inrush at turn on )

I agree 1n400x there

 

[LeeYoo]

Phantom protection speed.
Interesting forward (not reverse) recovery test of 4004 vs. 4007 diodes.
Seems the 4004 turn-on time is much faster.
http://www.cliftonlaboratories.com/diode_turn-on_time.htm#1N4001

 

[bruce0]

Very interesting article, but maybe you linked the wrong one.

This article says the turn on time for the diodes (used for relay coil snubbing) is almost the same for 4005 and 4007.  The focus seems to be on the forward release time which delays relay turnoff by 8ms. 

This doesn't seem to bear on the purpose of D7 and D8 since they will always be forward biased. 

Unless I am confused?

 

[LeeYoo]

Haha. Had already forgotten about D7/D8.

First line says phantom protection speed. The input protection diodes. I should have been more clearly here.
Stumbled upon this article while looking for info about how fast the phantom protection would work in case of a fault.
Normal diodes vs. zeners, schottky diodes etc.
Diode turn-on time is usually not found in datasheets. Just the turn-off time.
That link tested 4001-4s faster than 4005-7s.
As fast as a schottky diode.
Leo..

 

[bruce0]

Have you read the phantom menace returns article? 

If not it is really great.

Very detailed analysis of how to deal with the multiple kinds of phantom power faults (hot plugging, hot plugging one pin first, shorting one hot pin, shorting both, etc.)and all of the concomitant current flows (coupling caps discharging, flows through the preamp, all sorts of paths).

In the end they make some detailed recommendations.  They do use fast switching diodes in some places, but also identify areas where beefier ones are called for.

(they call for using GP glass passivated diodes as the input clamp between the rails because in general they have lower leakage, and leakage there inserts noise into the input lines of the preamp).

Your link was a very interesting article, thanks.

 

[LeeYoo]

Yes, I have read that article, and many others, before designing those 4-mic pre circuit boards.
THAT_Analog_Secrets_Your_Mother_Never_Told_You.pdf
THAT_More_Analog_Secrets.pdf
AES129_Designing_Mic_Preamps.pdf
AES5335_48V_Phantom_Menace.pdf
AES7909_48V_Phantom_Menace_Returns.pdf
Op_Amps_in_Line_Driver_and_Receiver_Circuits_P2.pdf
SSM2017 design notes.pdf
And many other datasheets of all the different manufacturers.
Leo..