Building ruffrecords' Universal EQ

[ramshackles]

So I'm doing this ambitious build of making a twin channel universal EQ with tube make up gain. Lets go!

I've got a number of boards from Ian, which I think should be enough to build the EQ:

2 * Universal EQ boards
1 Twin Line Amp board for makeup gain
3 * Lorlin switch mounting board
1 HT supply board.

Universal EQ
I'm starting by ignoring the power and gain sections for the moment and tackling the complex business of the EQ board. The universal EQ is seven passive filter sections on one board, which can be connected up however you like. There are four RC filter sections and three RCL filter sections. To read more about the possibilities, check out the original design thread:
http://www.groupdiy.com/index.php?topic=50484.0

There are 3 documents you need to read to have a hope of building the EQ boards:
Info on configuring a board and a 'standard' configuration using all sections:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/UniversalEQ/UNIEQConfiguration.pdf

How to calculate values for the RC filters:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/UniversalEQ/RCcalcs.pdf

How to calculate values for the RCL filters:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/UniversalEQ/PCB/InductorCalcs.pdf

This project is a good step up from simply populating PCB's and linking them all up as it requires a lot more preparation and a little more understanding of schematics. A great learning exercise!

Configuration
The universal EQ can be configured pretty much how you want. At the end of the configuration document, Ian describes how to create some classic configurations like Pultec, 4 band Pultec and Helios 69.
The schematic drawn on page 3 shows the EQ wired up with all sections in use. 7 filter sections! Sounds exciting and if I have understood correctly, it is entirely possible and the values given throughout the documents relate to configuring the EQ in this manner. Which seems the easiest way to proceed!


Examining the PCB
I started by taking a look over the PCB and trying to relate this to the info in the docs above, which feature simplified schematics of the relevant sections for each doc.
First thing to notice is the possible number of frequency positions. Each RC filter can cater for 8 positions, and each RCL filter can have up to 12.
Thats a total of 68 positions we could be playing with - 68 capacitor values to calculate!

The RCL sections have two alternate positions for the capacitors, which seems confusing at first but after looking carefully and reading through the thread, it seems there is good reason.

If I understand correctly:
[list type=decimal]
[*]The inductor position fits a Carnhill or ChrioN footprint, specifically the VTB9050 and the tap points for this inductor have traces to the first capacitor positions. The silkscreen labelling also refers to this inductor.
[*] You may wish to use other inductors though. To do this, you would mount the capacitors in the alternate positions (the unlabelled ones). The first pin of the alternate positions and the last pin of the original capacitor positions are linked, so you can put a jumper from the inductor tap to the last pin of the original capacitor positions.

[/list]
Here is a crude drawing of what I mean.



It also looks like only the second capacitor position is connected to the switch header, so if using the first capacitor position I must jumper across to the switch header.

There is also provision on the PCB for a pad and bypass switch, although I'll come to this later.
The only other important thing to notice is that there are no values for (almost) all of the components. This is because you will need to calculate them!!

Calculating component values
The documents I have linked to above give all the necessary information to calculate the C, R and L values. I have created a spreadsheet for myself to simplify things even further.
The first thing to do is decide upon resistor & potentiometer values. As I am wary of making mistakes and will be using Ians twin line amp for make up gain, I'll going to go with the standard configuration and values given by Ian.

The spreadsheet
I've created a spreadsheet so there is no need to faff around with doing calculations by hand and to allow me to play around with values without having to redo everything.
The spreadsheet is here:
https://docs.google.com/file/d/0B6LqEShkjOdRbm9EeFI0VWtTY00/edit?usp=sharing

For the RC filters, you simply add your desired corner frequency and it comes up with exact capacitor values. You can then add realistic capacitor values and see how this changes the frequency.

For the RCL filters, add your desired frequencies and 'Q' value and it comes up with the bandwith and exact inductance values required.
Then it gets a little tricky in terms of logistics. You will likely come up with a distinct inductance value for each frequency position. But inductors have at most 6 or 7 taps at predefined values.
So what you need to do, is take a look at inductor datasheets, like this one:
http://www.audiomaintenance.com/downloads/carnhill_design_guide.pdf

And choose inductors (that will fit on the PCB) that have the right range of taps for each filter section. You then need to put in the 'real' inductor value in the spreadsheet. The bandwidth will be recalculated and an exact capacitance value calculated. Again...there are only a certain number of capacitor values around, so you'll need to find the nearest match and put it in. The actual frequency will then be recalculated!!

So you can see, in theory, it is very easy to calculate the values required to get whatever frequency you want. But in practice, as there a limited number of C values readily available, and an even more limited number of inductance values, it will probably take a fair bit of playing around to get some 'real' component values coupled with some useful frequencies and bandwidths.

Phew! All the above is explained clearly in the spreadsheet.

So, I started by putting in values for the RC filters. As you can see from the spreadsheet, I started off with exact frequencies that relate to musical notes. There is even a musical note vs freq page on the second sheet.
But, I quickly got bored after realising that the limitations in capacitor values aren't going to allow for such precise frequency selection. So I quickly filled in the rest with a broad range of frequencies, trying to keep most detail in the mid-range (where most music happens!) while still providing a broad range of high cuts/boosts.
Luckily, with 7 filter sections this isn't so hard to do.

I filled in all 8 values for the RC filters, but it seems that 8 position switches aren't so easy to come by. Six position switches are readily made, so I may need to revisit this section.


I have not yet finished filling in the RLC section.
One thing to note is that each RLC section is in parallel with each other, so if frequencies overlap, instead of the boost/cut adding together, it will remain but be sharper (I hope I'm not making a mistake here.). What we should take from this is that you should not make the frequencies overlap!

Questions
There were a couple of things that I am struggling to get my head around in the pdf's provided for the project. These are mainly to do with the RLC sections:

[list type=decimal]
[*]R4 is referred to in the InductorCalc pdf as having a value of 2K2. I am having trouble relating this resistor to anything on the original schmatic or PCB. Is it Rcut1 and Rcut2? This is fairly critical as the inductor calculations are based on the sum of R4 and the source resistance.
[*]I am assuming that the source resistance of 150R as quoted in the InductorCalcs pdf would be the source resistance presented to the RLC filter stage from the previous stage, if wired in the standard configuration given by the UNIEQconfiguration PDF?
[*]As I uderstand it, the role of Rfix is to allow the parallel sections to still provide boost if one (or two) of them are switched to 'cut' mode. Does this mean that R6 should take the same value as VR6/VR8 (47K) ?
[*]The value of Rshelf is given as 220R if VR2 is 2K2. I find this gives a shelving frequency a decade above the cut. For high cut this seems a bit odd to me as the shelf will be largely inaudible (e.g. for a cut at 5K, the shelf is at 50K)...making it practically not much different to a non-shelving type. Have I got this wrong? Is there anything to be concerned about if raising the value of Rshelf?
[/list]

I have thus far not mentioned much about the resistors in the project, as I am simply taking the values offered by Ian . But I'll give a run through of what each of them are for reference. I'll leave blanks when I am unsure!

Firstly, you will see the resistance 'Rs' mentioned a lot in the RCcalcs PDF. If you don't read carefully first time round (like me), you will find yourself scouring the PCB for Rs (like me). It cannot be found. That is because Rs refers to the total resistance of the boost/cut pots in the RLC section. This forms one side of a potential divider, which is what makes the RC filters work.
With the value of the boost/cut pots - VR4, VR6, VR8 - at 47K, the total resistance is 15.7K - they are in parallel.

Now the rest:

• VR1 220K: for Low Shelf Cut
• VR2 2K2: Boost/Cut for High Shelf
• VR3 22K:  for Low Boost
• VR4 47K: for High boost
• VR5 4K7: Q for high boost
• VR6 47K: Boost/Cut for Low Mid
• VR7 4K7: Q for low mid
• VR8 47K: Boost/Cut for High mid
• VR9 4K7: Q for High mid
• Rshelf 220R(?): Turns the relevant RC filters into shelving types.
• Rbump 27K: Raises the low shelf cut frequency so that along with the low boost, the classic pultec 'bump' is obtained
• Rp: Entirely optional and would serve to change the non-shelving Low Cut into a shelving type (like Rshelf). If you are using the Low Shelf section, this is probably redundant
• Rfix: unsure of value
• Rcut: unsure of value

So that is the start. Hopefully a useful resource for other UEQ builders and even more hopefully, I might get pointed in the right direction when I make (inevitable) mistakes!

 

[ruffrecords]

That is one impressive post and an incredible spreadsheet. I think you have really caught the essence of what the Universal EQ is all about. A few points and some answers to your questions.

You have noticed that there are positions for two caps in series on the RLC sections. There are two reasons for this. One is so that you can use one section as an extra RC section as I did in the Helios 69 EQ example. The other is so you can more easily tweak the capacitor values to get hard to buy or unobtainium value capacitors. For example, if you need 18nF but cannot get it you can use a 100nF in series with a 22nF to get it. Or you could use a pair of 100nF in series to get 50nF, or a pair of 22nF in series to get 11nF and so on.

R4 in the inductor calc document is usually VR2. I wrote this document first and before I had got my head around explaining the basic operation of the EQ. I think I did it best in the Conficuration doc. The thing is, R4 does'nt have to be VR2. It is simply convenient to make the basic pot divider as shown in the RC doc but you don't have to do it that way.

The 150R source impedance is what the EQ as a whole is assumed to be driven by. The idea is that the Q of RLC circuits is mostly determined by the EQ components and not what's driving it.

The idea of Rfix is to ensure that the base level attenuation does not change when an RLC section is flipped from boost to cut. So you are correct that the value of Rfix should be the same as that of the pot it is associated with. (this is covered in the Config doc)

Unfortunately the laws of physics dictate that for about 20dB of cut, the top cut shelf frequency must be 10 times the frequency at which cut starts. Shelving EQ is usually described by the shelf frequency not the cut frequency. If you check out most bass/treble controls you will see that the ones specified as so many dB at 10KHz do actually start to cut at 1KHz. You can raise the value of Rshelf but since the cut at the shelf is approximately pot value/Rshelf, you also reduce the cut at the shelf. The thing to remember is that as the pot is varied, the shelf frequency stays pretty much the same and the cut frequency moves. So a 2dB cut that shelves at 10KHz has a cut frequency much higher than a 10dB shelf at 10KHz.

Feel free to ask questions and I will do my best to answer them.

Cheers

Ian

 

[bruno2000]

I would love to take a look at the spreadsheet, but would like it in Excel.  Anybody out there converted it?
Thanks!
Best,
Bruno2000

 

[ramshackles]

here it is:
https://docs.google.com/file/d/0B6LqEShkjOdRcTlXUFBfblJkMmc/edit?usp=sharing

 

[ramshackles]

Im not sure if google drive will allow you to download, so let me know and Ill put it elsewhere. I'll be updating it shortly aswell, to allow for different desired Q's in each of the RLC filter sections

 

[bruno2000]

Got it. Thank you VERY much!
Best,
Bruno2000

 

[ramshackles]

Spreadsheet has been updated with some minor changes.
I started to lay out a front panel to see if I could squeeze it all in to a 2U case, but it is impossible for two channels, so 3U it is.
With the Lorlin switches needing a 10mm mounting hole, the knobs I would like to use (Elma winged knobs with skirts) will have a total diameter of around 19mm.

I thought that the low shelf RC filters (boost and cut) could have their frequencies controlled by the same switch, (like a pultec), as I would probably use it mainly for that pultec bump. I presume there is no problem with running 2 wires from each switch position?
This will also save a little bit of space on the front panel.

A quick layout for the 3U version is here:
https://docs.google.com/file/d/0B6LqEShkjOdRZHA1R3I4NGM5ZVE/edit?usp=sharing

I have calculated all my inductor and capacitor values now, although I hadn't thought of using the 2nd capacitor position to make different values, so I will tweak my spreadsheet further and try to get closer to the exact values.
Although I suppose that using the first capacitor position would oblige me to have the 'standard' inductor connections for a VTB9050 as the PCB tracks are connected up?

I have come up with capacitor/inductance pairs using the following inductors: VTB9054; VTB9043 and VTB9042, allowing the 3 RLC filters to cover a frequency range from 300Hz - 25k Hz. I've not checked the PCB yet to see whether this is actually possible.

As this part of the unit is passive, do I need to consider the voltage rating of capacitors much?

I am looking at potentiometers for VR1 - 9. I am wondering how much difference in quality is there between a metal foil pot with 20% tolerance and a cermet pot with 10% tolerance. There is a ten-fold difference in price...

Next, the twin-line amp. It would be normal to put the gain before the EQ section?
I have not thought much about the outputs...is running unbalanced out fine, or should I be putting transformers or even an amp?

 

[ruffrecords]

A note about tweaking caps to exact values. It is not really necessary simply because both the cap and the inductor will each have tolerances of around 5%. The series capacitor technique is most useful for getting unusual values rather than obtaining high accuracy.

Capacitors of 50V working voltage or more will be fine. As they are not polarised they will take +-50 volts peak to peak which is about +33dBu. Your input transformer is likely to saturate first.

With all passive EQ is is normal to have the gain after the EQ. The normal signal flow would be:

Input Transformer ---> EQ  ---> Make Up Gain --->  Output transformer

The Twin Line amp holds the input transformer and the make up gain. EQ and output transformer are external to it. The Twin Line Amp will drive a VTB2291 transformer wired 2400:600.

It's a pity the tolerance on pots is so poor. I don't think it is worth the extra money for cermets. The exact value is not too important for the mid boost/cut circuits because they act like pots. It is more important for the RC circuits where the pot is one of the R values. Depsite their poor tolerance, I have found batches of pots from the same manufacturer to be remarkably consistent and generally on the low side. If you want really accurate resistor values then you are into switched versions.

Cheers

Ian

 

[ramshackles]

Thanks a lot. Yes I had originally put the twin line amp after the EQ, but seeing the tinw line amp had input transformers I confused myself. Now I see they are seperate to the rest of the circuit...
Looking at the prices of potentiometers, making switched gains are not too pricey.

I started putting together a BOM (mouser seems to have the best capacitor selection) but then realised I have not accounted for being able to bypass each section. At this rate, if the EQ is on there will always at the very least be the HPF on.
Is it best (or even possible) to add a bypass switch for each section, or would leaving frequency switch position empty be ok?

 

[ruffrecords]

Thanks a lot. Yes I had originally put the twin line amp after the EQ, but seeing the tinw line amp had input transformers I confused myself. Now I see they are seperate to the rest of the circuit...
Looking at the prices of potentiometers, making switched gains are not too pricey.

I started putting together a BOM (mouser seems to have the best capacitor selection) but then realised I have not accounted for being able to bypass each section. At this rate, if the EQ is on there will always at the very least be the HPF on.
Is it best (or even possible) to add a bypass switch for each section, or would leaving frequency switch position empty be ok?

It is not usual to have a bypass per section. All the sections but the HPF can be set to flat simply by turning their respective pots to zero (anti-clockwise). You would usually have one HPF switch position labelled flat and fit a wire link there rather than a cap. That way you can set the entire EQ to flat and it will be. If you can be bothered to do this then don't need a bypass switch.

Cheers

Ian

 

[ramshackles]

Ok so lose one cap position on the HPF.
If I have understood correctly, there is provision for a bypass switch for the whole EQ, aswell as a pad (on the section of the PCB marked PAD).

I did a quick tot up of all the parts the other day, and it will be around £600 for everything. So - it will be a long build using the remains of the monthly pay check

I'm starting on the twin line amp, as this will be a relatively cheaper segment (you wouldn't think it would be, as the EQ is mainly capacitors...but it is quite a lot of capacitors, plus 3 inductors per board, lots of rotary switches and pots, which shoots up the value).

The twin line amp also seems to be exceedingly easy to build, thanks to this lovely document:
http://www.ianbell.ukfsn.org/EzTubeMixer/docs/EzTubeMixer/twinlineamp/TwinLineAmpConfig.pdf

watfordvalves.com seemed to be the cheapest tube source in the UK, and the input transformers are readily available from sowter.

Quick question - on the document above, the power rating is only specified for 1 resistor, so I assume it doesn't really matter for the rest?

 

[ruffrecords]

Quick question - on the document above, the power rating is only specified for 1 resistor, so I assume it doesn't really matter for the rest?

Yes, 0.25W for everything except RHT which should be a 1W type.

How do you get on with Watford Valves? Their prices look good but they never tell you the shipping cost even at the checkout. Just makes me a tad supicious.

Cheers

Ian

 

[ramshackles]

Yes, I was a little suspicious, but I went for it. VAT should add another £7 to my order. They say minimum postage charge is £4 for the UK and I would be a tad p*ssed if they went over £5-6 as it is just 3 tubes.

Anyway, I made the order 3 days ago and apart from a confirmation email, nothing has happened. Nothing left my account either.

 

[ruffrecords]

Yes, I was a little suspicious, but I went for it. VAT should add another £7 to my order. They say minimum postage charge is £4 for the UK and I would be a tad p*ssed if they went over £5-6 as it is just 3 tubes.

Anyway, I made the order 3 days ago and apart from a confirmation email, nothing has happened. Nothing left my account either.

Let me know what the end up charging.

Another place I use a lot is hotrox in nottingham:

http://www.hotroxuk.com/

Cheers

Ian

 

[ramshackles]

I was writing about giving watforvalves the weekend before complaining (as thats all Im waiting for) and it turned up! They charged £4 for p&p and VAT was £7.40. They were packed in 2 jiffy bags and seem to be in just great condition.

I am a little scared about pushing the tube into the tube pins - I dont want to push down too hard! Is it normal that the legs of the tube sink about 3mm into the tube pins? That leaves another ~3mm exposed between the base of the tube and the tube pins...

How do people normally solder stuff that won't stay in place when you turn the board upside down - tube pins?! They were a nightmare.

So, as I wait for some more beans to buy another batch of components, I was thinking about another variation I'd like to build, using just the 'hi-mid' and 'lo-mid' sections...and perhaps a 3rd duplicate of it; so 3 bell filters, with fairly narrow-Q (say, 4-12 or something).
If I were to take just the 2 centre sections (lo-mid, hi-mid) and duplicate one, to give 3 cut/boost bell filters all in parallel, I suppose I would then need to attach a resistor to ground in place of VR2, to make up the other arm of the potential divider?
I imagine this then going to a make up stage (not tube) on the same circuit board and then I could fit the whole thing a 3U 500-series style module or something.

 

[ruffrecords]

I am a little scared about pushing the tube into the tube pins - I dont want to push down too hard! Is it normal that the legs of the tube sink about 3mm into the tube pins? That leaves another ~3mm exposed between the base of the tube and the tube pins...

How do people normally solder stuff that won't stay in place when you turn the board upside down - tube pins?! They were a nightmare.

There are two types of tube pins you can buy; one plain and one with a knurled shaft. You should use the knurled ones = Farnell part 149319. These should be a tight push fit into the PCB and should very definitely stay in place.

The tubes do not go all the way into the pins. There is usually a couple or three mm showing. When I first used them I was convinced the tubes  were just going to fall out but they don't.

Cheers

Ian

 

[ruffrecords]

So, as I wait for some more beans to buy another batch of components, I was thinking about another variation I'd like to build, using just the 'hi-mid' and 'lo-mid' sections...and perhaps a 3rd duplicate of it; so 3 bell filters, with fairly narrow-Q (say, 4-12 or something).
If I were to take just the 2 centre sections (lo-mid, hi-mid) and duplicate one, to give 3 cut/boost bell filters all in parallel, I suppose I would then need to attach a resistor to ground in place of VR2, to make up the other arm of the potential divider?
I imagine this then going to a make up stage (not tube) on the same circuit board and then I could fit the whole thing a 3U 500-series style module or something.

Yes, that's basically all you need to do. The three pots in parallel from the top arm and you just need a fixed resistor for the bottom Cheers

Ian.

 

[Jeremy.Starseed]

Glad to have found this thread, and the brilliant Excel spreadsheet, commentary etc.

Looking at the spreadsheet, it makes sense why I was apprehensive about getting started with the project, as it will be my first EQ build.

One thing I am wondering about is which switches and potentiometers to choose? Making stepped gains for 0.5 or 1 dB steps would be ideal, is that dependent on VR?  Been awhile since I had someone explain EQ circuits to me.

Thanks!

 

[ruffrecords]

I suggest you start by reading the Universal EQ documentation which begins with a basic explanation of  how the EQ works:

https://drive.google.com/file/d/0B_n67A1hN3qtSElVR1ZZSjJpSEE/view?usp=sharing

Cheers

ian