Idea for a dedicated Rhodes preamp, need help !

[PermO]

So, back to the circuit now...

Last night I spend some time with the passive pick-up emulation and tone stack and came to the coclusion that...

It sucks

It's six different shades of mud on the dial, and I have never been in the situation like "there's way too much sparkle, this needs some mud !"
So out it went and I tried some different filters like a passive high pass, it works, but I don't like how it sounds.
It does not make sense to have a high quality high impedance fet input with a bunch of passive junk strapped behind it, all it will do is kill your sound and suck the life out.

So, Douglas Self to the rescue... it makes way more sense to have a baxandall type EQ between the two opamp stages, utilizing the high impedance fet input, and use the pots for tone control and use the Elma rotary as a stepped attenuator at the output.
I guess this will sound so much better, so that's what I'll do.

My plan is to use this circuit;

I love how elegant and simple this is...

I also listened to the small Beyerdynamic transformer and I liked it, so that will stay as I like the idea of galvanic isolation on the auxilary output, it will do no harm and I have plenty, it will add zero cost for me. I think it adds a nice upgrade to the overall design.

I'll go back to the software now and draw a new schematic for as long as the pinched nerve in my neck allows me to do that.

 

[MidnightArrakis]

Nice job on the transformer !

But I don't have the file for the box that I have.

The thing is, creating this thing in Fusion360 for me will take way more effort than to just build it in the real world.
Recreating all parts on my desk will be a tremendous effort

[Nice job on the transformer!] -- THANKS!!! And.....since you have mentioned that you have 50 of these things and now I see that these are no longer made.....I am guessing that you must be sitting on a -- GOLD MINE -- worth of valuable transformers!!! These transformers are probably a perfect fit for some "500-Series" mic-preamp project, huh?

[But I don't have the file for the box that I have] -- Who makes it and what's its Part Number?

[Recreating all parts on my desk will be a tremendous effort] -- All of the parts for your project are more than likely downloadable from someone somewhere!!! You just have to know where to look.

[creating this thing in Fusion360 for me will take way more effort than to just build it in the real world] -- You are imagining working in a CAD-program to be way more difficult than it really is!!! As an example, creating this module wasn't difficult at all, despite how it may come across by how it looks:









[Recreating all parts on my desk will be a tremendous effort] -- You only need to recreate that which doesn't exist, like your TR/BV transformers, which was not at all difficult to do!!! Otherwise, you just download the parts, as they have already been created either by the companies that manufacture them or by CAD-fanatics that enjoy creating cool-looking parts. There are several websites online that specifically host -- TONS -- of CAD-models of practically everything imaginable that were made by other people like myself!!! Again.....you just have to know where to look.

>> HAPPY TRAILS TO YOU!! <<

/

 

[PermO]

I also have the 1200 ohm input variant TR / BV 310 001 061.

And a bunch of Lundahl 1527 and some unknown transformers from a radio installation that sound amazingly euphoric.

I worked a couple months in a broadcast workshop, assisting builds, extra hands... and the last day on the job I found a huge box of audio transformers on my desk with a note "Thank You !"

You get called to design these things...

I get called to build these inhouse designed small runs, as these boxes just keep sitting there when the designwork is done.

 

[dmp]

Nice project!

Looking at your schematic, if you have DC coming in, IMO the filtering caps is overkill. You could have the 0.5 ohm into the 220uF and be good. The resistors would serve as fuses. Would free up space. Maybe I missed it but why isn't there a FX return and blend?

A tip, I use net flags in EasyADA which cleans up you schematic a lot and makes it easy to avoid drawing mistakes, like mixing up the V+ and V- of ICs (which I did recently on a plate reverb driver PCB). Instead of drawing a line from your -18v supply, put a net flag there, label it Vnn, then at the negative pin of each IC put the same net flag.

My experience has been the EasyADA autorouter is pretty bad. Few things I do: 1) route important connections manually first to make them short and clean, then fill in the rest with the autorouter, 2) go over the whole thing and switch traces on layers, delete unnecessary vias, etc... to clean it up. The autorouter will sometimes add vias that are totally unnecessary, just based on the iterating algorithm (you have some examples of this, one by the U1 decoupling cap). 3) set your design rules before autorouting. You can make your power nets thicker traces.Having the nets labelled Vnn, Gnd, Vpp makes this easy. Increase the clearance on all nets to prohibit the traces from going between the smd pads. This is a small enough PCB you could just route it manually, but figuring out the autorouter is nice for bigger circuits.

You could change the BP footprints to be correct if you care to, currently they are indicating a polarized electrolytic. Figuring out how to modify a footprint is a little cryptic in EasyADA, but start from the footprint manager in the schematic tools menu, create a copy of the footprint in your personal library, edit it, then chnage C2 & C3 in the footprint editor to your new BP footprint. This is also how you would increase the pad size or hole size on a component if needed. You can overrule in the PCB drawing also, but then if you re-import you lose all your changes and have to do it again.

 

[PermO]

FX return and blend have crossed my mind indeed...

In this case that would mean, extra opamp, extra pot, extra jack... that would need a total redesign of everything.

But the plan is to do a 1HE rack version at some point, mains powered, dual ch. VU meters, FX send and return and a clean -9V output for pedal power, the ultimate instrument preamp.

I don't think there's anything out there with all these features.

 

[PermO]

Updated schematic;


Ok, here's some things I'm not so sure about...

THAT input and EQ section are driven from the same source, do I need to decouple the EQ section with a cap ?
Would this setup have trouble driving the EQ section, would it need buffering ?
Would the 2134 output drive the output transformer, would this need buffering ?

I really like where this is going so far.

(edit) I am aware this EQ circuit will flip my phase, I plan to correct that on the primary winding of the output transformer.

 

[MidnightArrakis]

Found a cheap replacement for the rather expensive Elma rotary switch;

Looks like it's about the same size, provides 8 gain steps.

https://nl.aliexpress.com/item/4000...il_item.3.4c916d76q9tXOK&gatewayAdapt=glo2nld

[Here's the -- same thing, only different -- should you desire to have the switches on the "side" of your enclosure, instead of on the "top":

https://www.aliexpress.us/item/3256...4ad3553-60af-47b2-8b99-a34439870778&_t=gps-id

/

 

[PermO]

Yes, I have some of these, those aren't too bad actually... in general I'm not a huge fan of Chinese switches.
Yes they are cheap, but you can always see and feel why that is.

I have used one as a impedance selector in the first build of this circuit, the proto-preamp.

It all depends on... what do we want to design ?

A GDIY project ? ...
We do a circuit board and frontpanel design and provide a BOM and keep cost as low as possible and focus on easy assembly, parts that are easy to obtain everywhere, and cost effectiveness

A commercial Unit ? ...
We do a complete design, so manufacturing can be outsourced ? Who's going to pay for all that ? I'm not an investor and don't have money to trow at a project like that.
Go for the bottom, get the cheapest Chinese tin can and switches and electronic parts, total productioncost of 1 unit be 30,- retailprice 399,- ...
Yes, that could be commercialy viable, but I hate that mentallity, producing crappy junk that can go straight to the landfill just to make a profit is not my thing.

A commercial unit that we will produce ourselves ?
High end, the best of everything, not cost effective but fingerlicking good.
Build two, sell them, build 4, sell them, build 8 sell them... etc ?

I don't know.

 

[PermO]

Update;

Cleaned up to give clearer picture of what's the plan...

I talked to a friend of mine and he told me that driving the output switch into the transformer from the opamp would probably not go well, so a buffer is needed.
I added a fet as I have used IRF510 for this in the past with good results, but those are rather big and a small Jfet should do, I have one NTE312, that could work but if anyone has a suggestion for a small and widely available part for this purpose, that sounds good, I would like to know...

Driving the EQ section and THAT chip from the same opamp output like this should be fine I guess ?

 

[PermO]

Updated brainfarts;

 

[PermO]

So to see if this aux output configuration was a good idea...

I build it;

And this time I was not disappointed.

I build the whole thing exept the THAT driver output, as I allready know how that fares.

I really like how this EQ turned out, it's a very usefull feature and it sounds pretty good.
Driving the transformer like this seems to work fine, I guess any PNP that can handle 36V will do.
The transformer will crap out on high levels crancking up the low end to extremes... wich is not a feature.
But in normal operation it handles just fine, I have not figured the compensation network on the secundairy yet, I'll need some more testing to see what's happening there.

Rhodes, bass guitar and synths all sound fine, I like where this is going, a very clean main output and a aux output with a little different character.

To be continued...

 

[PermO]

I have started the layout of my board.

This is my starting point, I'll need to solve some things using a via, but I might move things around to get a better layout.
As I'm doing it, I keep learning and finding new things, so I'll just keep on doing it.

As always any suggestions are welcome.

 

[MidnightArrakis]

I have started the layout of my board.

As always any suggestions are welcome.

As only as a "General Rule":
1) Use one side of the layout for horizontal routing and the other side of the layout for vertical routing. Doing this will initially alleviate getting yourself "stuck" with how to get a component routed. Of course, there are always "exceptions" to this rule, especially when you have two pads on the same net that are right next to one another. It wouldn't make any sense to change layers just because the net between those two pads goes off 90-degrees in its direction.

2) Always try to use the thickest track-width allowable in any given situation (within reason, of course). In other words, why use a 12- or 15-mil track when there's more than enough room for a 25- or 30-mil track??? No matter what, YOU are paying for the copper on the board, so you might as well use as much of it as you can, instead of etching it off and having it wash down the drain!!!

Here's some PCB-routing information that should help you out in your endeavors:

https://cdn.imagearchive.com/groupd...51f5f0cfff067e94a9ddb06a8ed754a3f79046d6aec32

https://cdn.imagearchive.com/groupd...6a6743f99024c7b7ac4d6cf76fe5a70bb2d625156c57a

https://cdn.imagearchive.com/groupd...19f27558da38fba8de63efcb5d1d70c66d7fa61371ac6

https://cdn.imagearchive.com/groupd...97b7e7022e8e902a1561a057cc7a02dea099b72533e15


/

 

[MidnightArrakis]

Here are some YouTube PCB-design videos:











GOOD LUCK!!!

/

 

[PermO]

There was some very insightfull advice there...

If you run into trouble designing a dual layer board, ask yourself "why two layers ?"

Good question, as there's 4 things that cannot cross and need to go everywhere on the board, Vcc, Vnn, GND and Signal.
So now I went with a four layer board, used bottom layer for ground, top layer for signal and the inner layers for Vnn and Vcc.

It looks so much better now.

 

[PermO]

So here's what I have now.

Ground;

Vcc;

Vnn;

Signal;

I think this should work.
For one thing, it's much easyer now to check all the connections with the schematic.
Don't know if a 100% ground poor would improve things here ?
Or any other conflicting issues.

 

[dmp]

there's 4 things that cannot cross and need to go everywhere on the board

I do 2 layer boards with vias. I think of a PCB as an extension of a PTP layout. I know a PTP layout can make a excellent audio circuit. Would you ever try to lay a copper sheet over a PTP layout? i.e. a ground pour. The enclosure should provide the shielding, I like to direct the ground path instead of making a big pour for it.

My principles are to make the signal traces as direct and clean as possible - cross other traces at right angles, keep differential pairs together, etc... then the power and ground I don't really care as much - vias are fine. Place decoupling caps near chips - make the different grounds (audio, power, relay, etc) separate with a jumper (I put in a 0 ohm resistor) and only connect them at the power entry on the board. Having traces parallel and aligned on different layers is something I would avoid on principle to avoid interactions. Not sure what the capacitance would be but you don't want crosstalk or signal bleed through a path like that.

You can look up trace width for amperage demand - anything over 1mm is probably completely unnecessary.

The mistakes I make are related to the footprints. I had a pot backwards recently. I also recently had a board made in EasyADA where the decoupling caps all were disconnected from the ground because of a bad connection in the schematic editor. They make it really easy to have a bad connection.

 

[MidnightArrakis]

[Don't know if a 100% ground poor would improve things here?] -- "poor" = > pour <, in this case.

[Or any other conflicting issues] -- At least on "this side of the pond", it is highly frowned upon to place the component REF DES (i.e., Reference Designator = R3, C9, F1, L2, Q1, etc.) -- INSIDE -- the component silkscreen!!! The general reasoning is.....because once the physical part is placed onto the board and it covers over the REF DES, then later on when a technician is looking for a component, -- THEY CAN'T FIND IT -- because it's blocked from view. But, you folks over there on "the other side of the pond" tend to do things quite differently than us "heathens" do over here!!!

This seemingly "unimportant" detail is actually rather important because placing all of the REF DES's outside of the component silkscreen will eat up space on the TOPSIDE of the board, so then your components placement will be directly affected by -- how and where -- the REF DES's are placed themselves. In other words, coming up with a good component placement becomes more of a challenge than you had originally thought it would!!! YA-A-AYYY!!!

One item that you may want to think about or, completely disregard and ignore, are the color assignments to the layers. Now.....since I happened to be involved with designing PCB's all during the time-frame of transitioning from manual hand-taping to using computers, everybody back then had to come up with some kind of scheme with how to identify the different layers and the use of colors was about the only way possible to do so. So.....in a general and loose type of sense, the following color scheme was devised and generally used. Should you be interested, I can later on explain -- WHY -- these colors are used. There actually is a "logic" behind all of this!!!

LAYER-1 -- TOP = RED
LAYER-2 -- GND = GREEN
LAYER-3 -- PWR = MAGENTA
LAYER-4 -- BOTTOM = BLUE
TOP SOLDERMASK = LIGHT GREEN
BOTTOM SOLDERMASK = DARK GREEN
TOP PASTEMASK = LIGHT GRAY
BOTTOM PASTEMASK = DARK GRAY
TOP SILKSCREEN = - WHITE - (If I actually make this text "WHITE", then it can't be seen)!!!
BOTTOM SILKSCREEN = CYAN
BOARD OUTLINE = YELLOW
DRILL HITS = ORANGE
GENERAL TEXT = GREEN
SPECIFIC TEXT = LAYER DEPENDENT

Then.....you want to try and avoid things like this:


----------↑↑↑↑----------------

That angle doohickey thing that is shown here is called an "Acid-Trap" in PCB-fabrication parlance!!! Meaning, when your board is being fabricated and going through the etching process, a minute (my - noot) amount of the etchant acid can become "trapped" down in this angle and/or underneath the copper itself in this angle. What happens is..... the copper right at this point is ever-so-slightly lifted-up and off the laminate material. Then, over a period of time, the track can work itself loose and possibly cause an intermittent circuit condition. The solution??? DON'T DO THIS!!!

/

 

[abbey road d enfer]

If you run into trouble designing a dual layer board, ask yourself "why two layers ?"

Because two-layer is exactly the same price as single layer with most small quantity PCB makers.

So now I went with a four layer board, used bottom layer for ground, top layer for signal and the inner layers for Vnn and Vcc.

Now this is overkill.

 

[abbey road d enfer]

Driving the transformer like this seems to work fine, I guess any PNP that can handle 36V will do.
The transformer will crap out on high levels crancking up the low end to extremes... wich is not a feature.

If this xfmr is really a 200r:200r, it is not the most adequate, nor the driver.
As you say, it's bound to crap out under high level bass.