Some EQ questions for the gurus

[raysolinski]

Let's start at the begining...I had a Urei 539 room eq that was on the fritz...so I had the idea to use the case, keep the pcb intact (removing the opamps and transistors) and build one of Fred Forssell's passive filter op amp eq's.....this unit has nice inductors and the power supply is pretty solid (+/- 17.5 volts with very little ripple)..I had one of Scott Liebers 2520's around so I built a line amp circuit, mounted it in the case and put the filter network in the feedback circuit as per Fred's design..http://www.forsselltech.com/downloads/schematics/EQ1A.PDF
I am doing 8 bands...What I noticed is the mid and high bands all funtion very well. The 2 lower bands function but there is definately less cut/boost and adjusting the low bands sometimes sends the 2520 into oscillation (motor boating)..I have been reading here like crazy and trying to wrap my head around this...the way the resistance of the pots would effect the q and amount of attenuation available...why the low bands seem to cut or boost less...hmmm so I thought I would seek the advice of the more experienced. For todays work I am going to limit it to 6 and then 4 bands...any thoughts would be appreciated  ;D

Cheers,
Ray

 

[jwhmca]

I had very similar results in trying my luck with a modded Pultec design... I found that the in and output impedance greatly affected the way the Hi/Low boosts would work

I think that if I had the input impedance wrong then the hi would stop working right and if I had the output impedance wrong the low would stop ( I was trying different in/out transformers of unknown Z...

Food for Z-Brain?

Just wait untill PRR or JR get a look at our Bumbling...!

 

[recnsci]

Ray,

Urei 539 na that Fred Forssells' Eq are quite different EQs. FF is so called "swinging input" EQ, while 539 is bandpass-in-feedback/feedforward-path. So, you would have to draw a schematic of hybrid you've made to be sure what is going on (I guess simplest approach would be to mark on Urei schematic what you have cutout/desoldered and how you've connected 2520).

In both type of EQs max boost/cut is deteremined by resistance in series with inductor and capacitor, more resistance - lower max boost/cut. In FF approach (unlike Urei) that series RLC network is connected to ground. Now, since you have low max boost/cut AND you have motorboating, my blind guess is that somewhere in there some ground path has significant resistance. That would cause less boost/cut and provide path for positive feedback which would cause motorboating. However, first post schematic so that rest of us know what we are debugging.

cheerz
urosh

 

[raysolinski]

The only thing I used from orginal 539 was the power supply (everything else was removed from the pcb except the filter network)..., output transformer and the case. I used selected inductor chains as they are...the path is resistor-cap-inductor-ground (the 539 pcb is labeled very well and the schematic makes it very clear) and placed them on the wipers of my 10k pots to ground as per Fred's design...the 2520 is just a 325 line amp circuit with feedback circuit designed by Fred that I mounted in the case...like I said the mid and upper bands are beautiful..it is the lower bands that are lacking in boost/cut..I think it is an impedance issue...reacting with inductor values and causing the bands to be shifted?..I think the motorboating may be caused by dc getting into the feedback loop as the lower bands are shelving types without a capacitor..I am going to add a high value blocking cap to feedback network to stop any dc... as a straight line amp it sounds very good..I am running it unbalanced in just like Fred's schematic at the moment but I really like that Urei output transformer!

Cheers,
Ray

 

[radardoug]

I take it you have a bunch of passive RLC eq sections in parallel on the front of your one opamp?
Simple answer, use another gain stage and put the low bands on that. The eq's are all interacting.
Also sounds like the high bands are quite high Q and therefore don't interfere so much with each other, and the low bands are lower Q.
Actually you should have just put better opamps in the 539, much quicker!

 

[raysolinski]

The filters are inside the feedback loop...not just in front of the amp like a passive design...the q's should be roughly the same as they are as designed in the 539...I am not a fan of chip opamps and love the sound of 325/312 amps and of course 550/560 eq's hence my little project..and this 539 needed alot of work...faders were bad and the gain section was really off...got it for cheap to use for this purpose as the inductors and output tranny are worth way more than the price of admission..when I realized the power supply was very solid after a recap I decided to use the case too and just made a new front panel..I will mess with it more tomorrow as I am working in the studio tonight and having a nice dram of single malt....

Ray

 

[shabtek]

I don't think I've seen a design, similar to the FF (1 opamp) mentioned, with more than 4 bands. Have you?
the 1/3 octave urei have buffers in the feedback.

 

[abbey road d enfer]

The filters are inside the feedback loop...not just in front of the amp like a passive design...the q's should be roughly the same as they are as designed in the 539...I am not a fan of chip opamps and love the sound of 325/312 amps and of course 550/560 eq's hence my little project..and this 539 needed alot of work...faders were bad and the gain section was really off...got it for cheap to use for this purpose as the inductors and output tranny are worth way more than the price of admission..when I realized the power supply was very solid after a recap I decided to use the case too and just made a new front panel..I will mess with it more tomorrow as I am working in the studio tonight and having a nice dram of single malt....

Ray

there's apparently some contradiction in your OP, where you put the filters in the feedback loop, when in fact the schematic dgm shows a "swinging input" design.

 

[raysolinski]

Well..it's just poor semantics on my part..the filter network is designed and implemented exactly as shown on Fred's schematic (swinging input)...his paper suggested that 10 bands were possible with this design..I decided to go with 8 but now see  it might be best to go with 4, 5 or 6..I will try them all today... hopefully

Ray

 

[abbey road d enfer]

Well it all depends on the level of performance you want to achieve in terms of noise in particular, and also the bandwidth of the filters.
I've seen this particular type used with as much as 27 1/3 octave bands, with RLC circuits (real inductors), although using two stages with interleaved frequencies is a better option.
The  swinging input circuit can become quite noisy if using simulated inductors (improperly called gyrators). In addition the noise gain of the opamp increases linearly with the number of sections, because of the combined resistance of all the faders in parallels. That's why commercial graphic EQ's use W-law faders.

 

[raysolinski]

The inductors are the real deal..I have to say this thing is really quiet, even with 6 or 8 bands..I figured out the q was too narrow with the resistors I used in the rcl networks..I calculated the bands for a Q of 1 (1 octave) and implemented them and they work much better...still having trouble with the low shelf..the high shelf works great..still a work in progress..thanks for all the tips so far!

Ray

 

[recnsci]

Ray,

just few more things. First, I guess u have tried this, but in case you haven't: take "suspected" inductor off pcb, and wrap it around line amp (breadboard style) just to check if it really an issue with inductor and not some issue with modified urei board.

Second thing, if you "only" take capacitor out of rlc network, whole thing acts quite different. First thing, with rlc resonant frequency stays the same [1/sqrt(L*C)] no matter what is amount of boost/cut. If you short cap, frequency goes off. Now "corner frequency" (aprox frequency where shelf starts to slope down/up toward unity gain of pass band) of low shelf is dependent of boost/cut (since corner freq is R/L where R is total resistance "seen" by inductor). As you cut/boost more, corner frequency gets lower. For instance, with 50H inductor (as in urei 50Hz band), 20K input and feedback resistors and 12dB of boost/cut, corner of shelf ends at around 15Hz at maximum boost/cut. So maybe the fact that action is quite different (and at lower freq than expected) to other bands is throwing you off? OTOH, that motorboating is still an issue (and I still suspect it's wiring/grounding issue)

 

[raysolinski]

Thanks Recnsci,
The indutors are "out" of the  539 circuit and just in my filter network..I am working on using different combinations of inductors and caps to get broader q factors on my bands..this is by no means meant to be a surgical eq..I got rid of the motorboating..I think my problem is exactly what you stated about the shelving..the unit is dead quiet and my goodness do the mid and high bands sound good...that nice lcr eq smoothness..I will work some more today and report back..thanks for teh help guys...almost there!

Cheers,
Ray