Barry Porter "Net EQ"

[frazzman]

And now with HF boost...

 

[Gustav]

Hi guys,
Almost finished my build so close I can taste it!
Just finished wiring on my right channel. All checks out except for an issue on HF band. Shelf is fine but peak isn't working for boost or cut., just stays at 0  - freq plot is flat.
I'm using stepped switches as per Harpos spreadsheet.
This channel is built identically to my (working) left channel. Any ideas ? If shelf works, where should I focus for peak issue ? I'm reading the right resistance on at the pcb so I'm know switch is wired correctly.
Thanks a lot

Did you figure this out?

Gustav

Hey Gustav,
Yes I did, thank you mate. Wrong resistor value somewhere.

Everything is pretty much working except for one overall anomaly, affecting both channels - even in bypass.
I'm observing some noise/ripple in the higher frequencies...Very prevalent with a HF boost. The plots from RMAA explain all....

It's quite audible, even though the unit is very quiet and hum free with no audio passing. With audio present  you can hear this 'ripple'.

Did you ground your pcbs ? I haven't done that and I am wondering if that's where my problem is?

Thanks a lot

I had ripple on a band which turned out to contain a cold solder in one of the cold solder (I realised at some point, the ripple only occurred when I had it over a specific frequency and checked the spots). That may be worth checking out!? Also, check for shorts/near shorts on the switch.

Remember, the shelf is separate from the bells, so you are switching to a totally different part of the circuit/different part of the switch between the two. If you see a problem in one and not the other,.

I am not sure what you mean about grounding the PCBs. I connected the ground points to chassis as marked in the schematic, of course, but maybe "grounding the PCBs" covers something I am not aware of.

Gustav

 

[frazzman]

Hi guys,
Almost finished my build so close I can taste it!
Just finished wiring on my right channel. All checks out except for an issue on HF band. Shelf is fine but peak isn't working for boost or cut., just stays at 0  - freq plot is flat.
I'm using stepped switches as per Harpos spreadsheet.
This channel is built identically to my (working) left channel. Any ideas ? If shelf works, where should I focus for peak issue ? I'm reading the right resistance on at the pcb so I'm know switch is wired correctly.
Thanks a lot

Did you figure this out?

Gustav

Hey Gustav,
Yes I did, thank you mate. Wrong resistor value somewhere.

Everything is pretty much working except for one overall anomaly, affecting both channels - even in bypass.
I'm observing some noise/ripple in the higher frequencies...Very prevalent with a HF boost. The plots from RMAA explain all....

It's quite audible, even though the unit is very quiet and hum free with no audio passing. With audio present  you can hear this 'ripple'.

Did you ground your pcbs ? I haven't done that and I am wondering if that's where my problem is?

Thanks a lot

I had ripple on a band which turned out to contain a cold solder in one of the cold solder (I realised at some point, the ripple only occurred when I had it over a specific frequency and checked the spots). That may be worth checking out!? Also, check for shorts/near shorts on the switch.

Remember, the shelf is separate from the bells, so you are switching to a totally different part of the circuit/different part of the switch between the two. If you see a problem in one and not the other,.

I am not sure what you mean about grounding the PCBs. I connected the ground points to chassis as marked in the schematic, of course, but maybe "grounding the PCBs" covers something I am not aware of.

Gustav

Thanks Gustav, I don't think the problem is specific to a particular band though as you can see the ripple even in bypass. Also both channels exhibit the problem so probably not a cold joint unless both channels have the same cold joint.

I'm guessing it's something like a wrong component on both boards.

As for grounding PCB, what I meant was - do you have 0V pin on molex XLR connector connected to chassis?

Another thought was maybe the issue is related to the psu. I'm using the Calrec psu here. This board is obviously common to both channels so that's a thought... (Since the problem affects both boards) Although the voltages from psu board are stable and correct.

Thanks again

EDIT: some sound clips to help further describe the problem:

- First clip is a  short unprocessed 1khz sine wave as a reference:
https://www.dropbox.com/s/uf18igqs1ui3lfu/sine_wav.mp3?dl=0

- Second clip is  then the exact same tone as above but this time through the net EQ in circuit but with no boost or cutting on any bands:
https://www.dropbox.com/s/r1lt4xuu2du5tq2/sine_wav_netEQ_BYP.mp3?dl=0

If you toggle between the two clips you can clearly here that something is a miss on the net EQ clip. It almost sounds like there is weird resonance in the high end, in line with what you can see in the frequency plots posted. The problem present in bypass is then of course exacerbated when doing a HF boost etc.

 

[biigniick]

thanks for the help, guys.  working on frequency calculations now.  i found harpo's spreadsheet

thanks,
- nick

 

[frazzman]

Sorry to keep hijacking this thread but just looking for input on this issue I am having.

As per the attached frequency response graph, there is a "ripple" in the high frequencies when the EQ is engaged. As it turns out this is not an issue when both channels are in bypass mode.

This "ripple" manifests itself audibly as a kind of resonant/oscillating sound in the high frequencies as soon as the channels are taken out of bypass - even with no cutting or boosting occurring on any bands.

It is less obvious on a whole mix but with a sine wave test signal it is quite dominant and is driving me crazy!

I have been through the input and output stages with a fine tooth comb but cannot find any wrong components.

Truly appreciate any pointers or theory's anyone may have that could assist!

 

[biigniick]

Frazzman,
It's hard to tell from the MP3 files, but it sounds like distortion of some kind. Maybe an impedance mismatch or level too hot on the record input?  Does the problem go away if you patch the eq out (disconnect the eq in and out and plug the wires together to capture just the signal looping through the cable)?

- nick

 

[frazzman]

Frazzman,
It's hard to tell from the MP3 files, but it sounds like distortion of some kind. Maybe an impedance mismatch or level too hot on the record input?  Does the problem go away if you patch the eq out (disconnect the eq in and out and plug the wires together to capture just the signal looping through the cable)?

- nick

Hey Nick,
Thanks for the reply. No issue when running the cables in a loop... Flat response...
The levels are adjusted to meet what RMAA wants, adjusted for -1 in the software.

As I mentioned there's a slight ripple in bypass on the frequency response curve but I can't hear a problem in bypass.

It seems as soon as the eq is engaged (out of bypass) that the problem occurs - even with no boost or cut on any bands.

The frequency response I provided was with a HF boost just to show the problem at it's worst. It seems to be exacerbated by boosting the high frequencies but it is still there regardless.

It's a total head f%#^!

Thanks for the reply mate.

Edit: FWIW - it's running into an RME fireface 800 (running at +4) incase someone has maybe encountered a similar problem , I doubt it's related but can't hurt to mention.

 

[ruairioflaherty]

Have you checked for oscillation with a scope and square wave combo?

What kind of THD figures are you getting in RMA for a loopback and for the unit?

Are you sure you are running the RMA sweeps at low enough levels to avoid clipping on output or input?

 

[frazzman]

Have you checked for oscillation with a scope and square wave combo?

What kind of THD figures are you getting in RMA for a loopback and for the unit?

Are you sure you are running the RMA sweeps at low enough levels to avoid clipping on output or input?

RMAA levels are definitely low enough... I wish they weren't as it would be an easy fix. Damn.

Anyway, the figures are pretty alarming:

FF800 with In/Out level set for -1.5dB which is what RMAA seems to like.

Loopback - THD: 0.0015
Net EQ - circuit in but no boost or cut on any band - THD: 0.763

I don't have a scope unfortunately so I'm behind the 8 ball for sure.

Cheers for your help. I will persevere... such is the joy or (un)joy of DIY!

 

[ruairioflaherty]

If that's -1.5dBFS it seems awfully high, any small response deviation or gain will cause your RMA rig to clip.  I'd be running that at at least -6dBFS, sure you'll have a 6dB higher noise floor but that's not a big deal IMO.

So it definitely looks like you have oscillation.  Have you built this with super trick op amps or the stock NE5532?

 

[kante1603]

If that's -1.5dBFS it seems awfully high, any small response deviation or gain will cause your RMA rig to clip.  I'd be running that at at least -6dBFS, sure you'll have a 6dB higher noise floor but that's not a big deal IMO.

Exactly what I thought!
Also in the general RME preferences check the i/o presets for proper high,4dBU,-10dBV or Low settings!These alter the reference points!


Good luck,


Udo.


Edit:For potentially important infos here on pages 43 & 45,tech. data start at page 82:


http://www.rme-audio.de/download/fface800_e.pdf


So e.g. the outputs are set to "high" means +19dBu for 0dBFS etc.,that would be on the pretty high side.

 

[frazzman]

Hey guys,

Cheers for helping out.

Just to clarify, when I referred to those input and output levels I was referring to the levels in RMAA as per the attached pic. I basically just adjust the in and out faders in totalmix until RMAA is happy. The FF800 is set for +4dbu.

I'm using standard ne5532s, I haven't substituted these for anything more elegant. Admittedly, I did buy a bulk lot of these from ebay - I'm hoping maybe they're bad quality but I doubt it as I've used them in other builds without issue.

I was thinking the fault must be in the input or output stages since the problem is present even with no boost or cut on any band....

 

[ruairioflaherty]

I would ignore the RMA suggested levels, that is sailing way to close to the wind for a device with any gain at any frequency.  That said I don't think that's your issue.

Do you have any local access to a buddy with a scope?  That's really the best tool to track down the issue as you can feed a square wave in and scope after the input stage, after the EQ bands, and again after the output stage.

It would probably also be a good idea to confirm that your PSU bypass caps on the NE5532 are all the right value, did you measure each one as it went in?

 

[frazzman]

I would ignore the RMA suggested levels, that is sailing way to close to the wind for a device with any gain at any frequency.  That said I don't think that's your issue.

Do you have any local access to a buddy with a scope?  That's really the best tool to track down the issue as you can feed a square wave in and scope after the input stage, after the EQ bands, and again after the output stage.

It would probably also be a good idea to confirm that your PSU bypass caps on the NE5532 are all the right value, did you measure each one as it went in?

Yep, all those 100nf caps are ok. I tried dropping my input and output levels drastically but it didn't affect the oscillation/distortion. It was prevalent even with low levels.

I don't know anyone who has a scope but I'm going to try and find one!

Just a thought - in regards to r20 and r26, these have been omitted as per the guidelines in Barry's doc, I take it that they don't need to be jumpered right ?

 

[ruairioflaherty]

Yep, all those 100nf caps are ok. I tried dropping my input and output levels drastically but it didn't affect the oscillation/distortion. It was prevalent even with low levels.

Cool.  Good practice to run that at lower levels especially if you are looking at EQ band sweeps with gain.

I don't know anyone who has a scope but I'm going to try and find one!

It will make light work of this.

Just a thought - in regards to r20 and r26, these have been omitted as per the guidelines in Barry's doc, I take it that they don't need to be jumpered right ?

Just took a quick look and both should be omitted completely and not jumpered.

Are your inputs and outputs to your RMA rig balanced?  If so perhaps you can make a cable to unbalance them one at a time by connecting pin 3 and ground.  Then you might be able to inject a balanced signal and probe the circuit with an unbalanced alligator clip - at several points along the circuit to see where the oscillation is creeping in?

 

[frazzman]

Yep, all those 100nf caps are ok. I tried dropping my input and output levels drastically but it didn't affect the oscillation/distortion. It was prevalent even with low levels.

Cool.  Good practice to run that at lower levels especially if you are looking at EQ band sweeps with gain.

I don't know anyone who has a scope but I'm going to try and find one!

It will make light work of this.

Just a thought - in regards to r20 and r26, these have been omitted as per the guidelines in Barry's doc, I take it that they don't need to be jumpered right ?

Just took a quick look and both should be omitted completely and not jumpered.

Are your inputs and outputs to your RMA rig balanced?  If so perhaps you can make a cable to unbalance them one at a time by connecting pin 3 and ground.  Then you might be able to inject a balanced signal and probe the circuit with an unbalanced alligator clip - at several points along the circuit to see where the oscillation is creeping in?

Good advice, thank you....

I'm regards to a signal probe - perhaps something like this :

http://web.archive.org/web/20050310014839/http://www.diyfactory.com/data/mbsignaltracing.htm

Mnats had a link on his site to this page and I made one of these probes but never had a need for it... (Until now)

 

[Harpo]

These -different project- posts from you might be related to your problem ...

Just have an issue with a the Calrec mini PSU board. Basically I etched one to use for another projector (2x net EQ). Anyway, I used alternate regulators to get a 18v supply. The only issue is on my 30VA toroid, the secondaries seem to be a bit hotter than normal. So I end up with about -/+ 20V rather than 18v.

20V at unload condition? That is OK in most cases, try that with some load and measure again, BTW if You use NE5532 op amp that's OK (+/-22V Max).

Thanks for the quick reply....yes at unload - actually closer to 21v

It turns out I'm an idiot and I accidentally ordered a 30va 2x25 instead of 2x18...
Will 7818 and 7918 regs play nice with this extra voltage ?
The net eq is jam packed with ne5532s so I'm glad to hear they will cope with the surplus voltage...

With your 2x25VAC secondaries unloaded and kept a secret mains transformer in front you operated the fixed voltage regulators above their abs.max. 35VDC input voltage (and as well maybe the storage caps), so these might be damaged and maybe still are with your now 2x18VAC transformer in front. If you once have connected this maybe faulty supply to your NetEq pcbs (from Gustav or self etched as well?), not all of the populated NE5532s might have survived. Just an idea.

 

[ruairioflaherty]

Good advice, thank you....

I'm regards to a signal probe - perhaps something like this :

http://web.archive.org/web/20050310014839/http://www.diyfactory.com/data/mbsignaltracing.htm

Mnats had a link on his site to this page and I made one of these probes but never had a need for it... (Until now)

I don't think you need to go that far.  Without a scope what you are looking to do is measure just parts of the circuit through RMA, first input stage, then after first eq bands, then after HPF, then after output stage.  It should help you find the area of the circuit that's oscillating.

To do that you'll need to go balanced out of and unbalanced in to your RMA.  It's crude but lacking a scope it could help narrow down the trouble spot.

 

[biigniick]

hi all,
i'm currently working on my first build and have a few questions about the harpo spreadsheet with frequency calculations and the resistor calculations for stepped switches.

the resistors and diagram are very clear and i understand the center-tapped pot replacement for the level switches.  we are making a string of resistors that all add together from the pole/filter to the boost/cut with the middle of the string attached to ground.  even steps make for even resistor values.  they all add up to 5k on each side. that all makes sense to me to replace 10k linear pots.  i order the values in the column "closest e96" of 500R and count for the quantity.  they are all the same and it makes sense for a linear pot they are the same value on every step.

my question is with the Q and frequency selectors.  they are also center tapped linear 10k pots, but the calculator does not give a string of similar values like the level calculator.  i'm using even steps, so shouldn't the values also be evenly spaced?  from what i see i'm order ing one 1k, one 953R, one 845R, etc...  does this seem right?  maybe i'm misunderstanding this?  am i going to use separate values for each step and wire them independently?  not like the string of resistors for the level switch?  they each seem to be 500R apart or so, could i wire them like the level switch with all the same 500R value and re-create the 10k pot?  are there noise implications?

i know that's a lot of questions, but that part of the spreadsheet hasn't quite clicked with me yet.

thanks,
any clarification would be nice.

- nick

 

[ruairioflaherty]

my question is with the Q and frequency selectors.  they are also center tapped linear 10k pots, but the calculator does not give a string of similar values like the level calculator.

Welcome Nick, good to have you here.

Just a small correction, the Q and frequency pots/rotaries are not center tapped.

Cheers,
Ruairi