LCR riaa's

the road not taken.
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey andrew,

can you upload the .asc, the image is a bit tough to read.

I originally started with the Tango and tweeked a few values, attempted to add in some parasitics and added the thrid time constant.

i did find an error in the file i uploaded, the resistance of L2 was off by a factor of 10. I corrected it, for future downloads, RL2 should be 14 not 140.

that might be some of the error you were seeing.

dave
James D
Posts: 15
Joined: Thu Mar 31, 2005 2:00 pm
Location: UK

Post by James D »

Nick,

This is what I get naively from your graphs...

+6dB at 1KHz looks like 300Hz at -10 on your graph... anyway that puts 0dB reference at -16 on your chart.

Then your noise floor (ignoring 50Hz pickup) is averaging around -136 so giving you a test noise floor of -120dB (nice)

Then the phono noise floor is at and average of -111 below 1KHZ and -126 above 1KHz

So without weighting your RIAA noise performance is -95dB below 1KHZ and - 110dB above 1KHz...

And that is bl**dy amazing! I not surprised you say that thing is quiet! Thats means you sig. noise is better than a CD player since CDs are specified relative to absolute peak level and you are relative to a 5cm/sec cut!

And that is with pentodes...

I also see the 50Hz that is there in the test set up but I don't see any psu components above the basic noise signature - nice one :)

I've probably got this all wrong but I can see where...

James
nickg
Posts: 29
Joined: Sat Mar 19, 2005 1:46 pm

Post by nickg »

Sorry brain fade, it was and is 300hz. I must get some better better cable and rerun the connection between the listening room and where the computer is to try and get rid of that 50hz. The card I am using is a m-audio 2496. I don't know what the noise floor in 24bit mode will work out, but its good enough for a simple chap like myself in 16bit.
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

hey andrew,

can you upload the .asc, the image is a bit tough to read.

dave
Hi Dave,

Sure, thanks for the update, I fixed the typo in the parasictics and have included the new .asc and the new graph.

I had a quick fiddle with the values and have it within 0.1db. Let me know what you think?

cheers,

-- Andrew
Attachments
Clipboard.jpg
Clipboard.jpg (199.46 KiB) Viewed 39111 times
LCR-RIAA.asc
(2.02 KiB) Downloaded 747 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

My .asc.

Post by dave slagle »

I just realized i uploaded the excel twice in my earlier post.

here is the .asc and the corrected excel.


dave
Attachments
LCR 6k(2).xls
here is the excel file again.
(97 KiB) Downloaded 763 times
tango variable 3rd constant.asc
everything scales except for R6. It must be adjusted manually
(4.84 KiB) Downloaded 744 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

I had a quick fiddle with the values and have it within 0.1db. Let me know what you think?
i think .1dB in spice is more accurate than any of the real world measurements that will follow.

I took my .asc and used 2.5 as the multiplier (to net a 2mic cap) and adjusted the following.

X=2.5
RL1=60*X
R6=67.7*x

and we are within 0.025dB :-)

i find it helpful to limit the plot area to like 10-40K so you can see things a little better.

dave
Attachments
this response looks ugly until you look at the scale....
this response looks ugly until you look at the scale....
Picture-1.png (6.51 KiB) Viewed 39094 times
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hey Dave,

0.025db we're splitting hairs 8)

I downloaded the new model and reworked to match your changes. Running your model I see much the same as you :D

If I have a matched source Z, rather than a matched load then I'm guessing I check my Rs to 1k5 and RL to 1Meg?

If so, I think this might be the source (pardon the pun) of my differences since my load will cause R6 to be different. Make sense?

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

i forgot that you were using a defined source. Spice should tell you what happens :-)

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

i forgot that you were using a defined source. Spice should tell you what happens :-)

dave
Yes, absolutely :) I'm happy :D

I guess I was just trying to reconcile what you were seeing versus what I saw. It's always good to close the loop.

So how do you fancy 'aving a go at winding a set of 1k5's?

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

the 1K5's should be pretty easy.

do you want to give me the inductor values and DCR's ?

I plan on putting taps at every 2% or so so you have some adjustabilty. Right now i am thinkng 9 settings total witht he 5th step being the target inductance.

we can talk in email or by the private message feature about the cost.

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Ok Guys, here's my best shot at the circuits.

LCR is sourced matched and terminated with a high resistance 3Meg3. Inductors are to Dave's spec. I tweaked some of the R to give a better RIAA in a source macthed system.

First image is the LCR itself.

Then a graph of freq response with the Aikido front end.

Then with the 6C45 Cascode front end.

Driver/second stage in both cases is a D3A.

I'm not completely happy with the cascode variant as yet, the 33mdb hump at high freq is bugging me.

The plan is to construct the amp in a modular fashion, that I can swap various bits around to try different configurations. I hope to add an LC only 1k5 RIAA into the mix for comparison as well.

cheers,

-- Andrew
Attachments
Aikido front end.
Aikido front end.
aikido.jpg (136.48 KiB) Viewed 38996 times
6C45 Cascode front end with load resistor of 1490R
6C45 Cascode front end with load resistor of 1490R
6c45.jpg (143.05 KiB) Viewed 38996 times
The LCR circuit itself.
The LCR circuit itself.
LCR.jpg (55.5 KiB) Viewed 38997 times
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave,

The 'as final as it will ever be' 1k5 LCR RIAA is attached.

cheers,

-- Andrew
Attachments
LCR-RIAA.asc
The 1k5 LCR RIAA circuit.
(2.03 KiB) Downloaded 800 times
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

So now I'm thinking "how am I going to build this?"

Just about every resistor combination needs two in series or two in parallel to make the correct values. Boy these were fun to order. I used TJ Morgan resistor calc to help get the best matches.

So here's a layout that can be used to produce a small PCB, about 2" by 3.5"

I should be able to get a "one off" of this built and it will make life much easier.

The two large gaps in the middle, either side of the GND, are for the 2uF caps - I may drill holes to mount cable ties or such like for the prototype. Dave's wonderful inductors will be mounted off PCB, direct onto the bread board for now.

cheers,

-- Andrew
Attachments
Clipboard.jpg
Clipboard.jpg (86.65 KiB) Viewed 38719 times
IslandPink
Posts: 56
Joined: Thu May 12, 2005 8:01 am
Location: Denbigh, North Wales

Source/Load

Post by IslandPink »

Andrew, Dave, Nick
Very interesting work .
I like the idea of 1.5K units as a compromise .
Now , to recap, I got lost towards the end of the discussion , are we saying this circuit, like the Tango or S&B units, will need -
(i) a well-defined 1.5K driving impedance ?
or
(ii) an unspecified driving impedance but a 1.5K terminating load ( eg. the grid resistor for stage 2 ) ?

..or something else ?

I still quite like the Pentode-driven circuit as per 2002 Thorsten, and just need to check on how this can be implemented .
Is any part of this likely to be canned or will it all be open and au-naturel-Slagle style ?

Mark
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Re: Source/Load

Post by Andrew »

Hi Mark,
I like the idea of 1.5K units as a compromise .
Now , to recap, I got lost towards the end of the discussion , are we saying this circuit, like the Tango or S&B units, will need -
(i) a well-defined 1.5K driving impedance ?
Spot on.. I have 3 driver circuits I'm working on; a 6C45 cascode, an Aikido and pentode.
or
(ii) an unspecified driving impedance but a 1.5K terminating load ( eg. the grid resistor for stage 2 ) ?
Yes, again, but you need a large coupling cap betwixt the first stage and the LCR, this has put me off this approach. Thosten had some clever ideas using two caps in a filter arrangement where the big cap only came into use at low freq and a smaller, more cost effective, high quality cap, could be used for the rest of the audio band.
Is any part of this likely to be canned or will it all be open and au-naturel-Slagle style ?
Over to you Dave.
I still quite like the Pentode-driven circuit as per 2002 Thorsten, and just need to check on how this can be implemented .
The advantage of 1k5 is a) smaller caps, b) standard caps sizes, c) slightly easier load, not much easier, but if your'e using a pentode/cascode you can give them a bit more load to work into 1k5 instead of 600R.

Eventually, I'll post my final circuits here, but I want to build them first. I'm worried someone might take my ramblings literally and build them and find they sound truly pants.

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Time for an update....

Here's a nice piccy of the Aikido front end, ready to go. I went a bit over the top with the boutique resistors, but its what I had laying round plus a few extras from PCX.

cheers,

-- Andrew
Attachments
Aikido front end for the 1k5 LCR
Aikido front end for the 1k5 LCR
aikido.jpg (93.99 KiB) Viewed 38622 times
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

mark wrote:Is any part of this likely to be canned or will it all be open and au-naturel-Slagle style ?
Au-naturel baby.

Canning them up should only happen after they have been listened too and tweaked. How could i (or you for that matter) possibly know what caps and resistors you like best?

keeping them naked also allows you to tweak the values to get the measured in circuit performance as close to the ideal as you want and of course it also allows you to include and play with the third time constant.



dave
James D
Posts: 15
Joined: Thu Mar 31, 2005 2:00 pm
Location: UK

Post by James D »

Neat pcb! Nice and clean lines. Bet you can't wait to try it all :)

James
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Dave's iron arrived on Friday, very nice it looks to....so I've been scheming, I basically want to get an experimental system going with what I have and see where I go; as you know my mind is yet to be made up on what is best topology and valve-wise.

So I have a 120v PSU that should get me going for a first pass to check everything out, this is quite a way short of my ideal but I will see where it takes me. Besides I might as well have a go while I'm waiting for AE to deliver the PSU iron.

First up will be the Aikido driving a C4S loaded triode connected D3A, that should get me running, the C4S makes the D3A a possibility, as for the Aikido I will have to see how it does at 120v.

I still have to wire Dave's iron into the resistors and will raid by box for caps until I can afford a nice matched 2uF pair.

I'll let you know.

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

So the arrival of Dave's iron has got he old grey matter going and I'm thinking where next once I have the test setup going. Now for me it takes time to think things through and ponder alternatives. I like to start early give myself time to work things out, do the maths, and order any parts.

My first idea was to use the Aikido as stage one. The main reason I went for the Aikido is that its low Zout, perfect to drive an LCR RIAA. But with 4 devices for a single stage it could be noisy; this has always cast a shadow over this idea. Now Thorsten's 600R LCR phono used an E810F pentode as a driver in place of the Aikido but with a 1k4-ish RIAA other pentodes can now be considered, most with a gm of >30 are fair game as well as a few cascodes. A 6C45 cascode can make gm of 30 easy. All of these should give a gain of around 40-45, roughly where Thorsten was, perhaps even better, with a 600R LCR and the E810F with a gm of 50.

But what about the second stage, it needs to be low Zout as well to drive the line but also provide gain, back to the Akido, perfect! An Aikido with a gain of between 20 and 25 will take me to the magic 40db figure for a MM stage and a great Zout 200R-ish. That also leaves plenty of valve choices for the gain segment of the Aikido anything wiuth a mu around 40 to 50.

But what else can we consider?

Last night I was reading an article about choke loaded triodes as a line driver for a phono; an intersting idea, it used a 5687. Now a meaty valve, with a choke load could be a perfect line driver, low-ish Zout but with gain as well and perhaps not as noisy as the Aikido say aim for 1k Zout perhaps less.

This would be a nice clean and simple design, two stages, two valves, with LCR between and approx 40db.

Dave whaddya think?

Can a choke load be optimized for low Zout and gain, I assume we're aiming to have low DCR and enough H to avoid low freq roll off.

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

This would be a nice clean and simple design, two stages, two valves, with LCR between and approx 40db.
that was the inspiration for the 7K5 LCR. Then two 6C45's would get you there. the problem i have with a low source to load ratio is you give up nearly 6dB of precious gain. At least a 600R loaded pentode feeding the 1K4 LCR would get you some of the gain back. (over 600R feeding a 600R LCR)

Can a choke load be optimized for low Zout and gain, I assume we're aiming to have low DCR and enough H to avoid low freq roll off.
not sure what you are asking here, the source impedance of a choke loaded stage will always be determined by the Tubes Rp. (above the lowest frequencies)

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave,

I didn't explain myself very well, sorry.

T's second variation of his amp used the E810F pentode with a 600R load on the pentode, rather than a 600R load on the RIAA, he used a 600R source impedance for his RIAA.

The difficulty with this approach is that he needed a very very high gm pentode to get any useful gain with a 600R load resistor. Assuming we adopt the same approach, the resistor for the pentode now needs to be 1k5-ish so we can use a lower gm pentode, 12GN7, DA3 and so on, should we wish. I used to have pentophobia, as a friend calls it, but high gm pentodes with low resistive loads should be very linear and do actually sound pretty good. There's also the cascode to try for those who really don't like pentodes, a 6C45 cascode gets into the gm > 30 range as well.

The result is a gain of around 40 to 50 or 4, possibly even closer to 5, after RIAA eq, so that a D3A triode connected or 6C45 could be used for the second stage as it only needs a gain of 20 to 25 or so for the magic 40db. However, stage 2 needs to drive the line as well, I was think a choke loaded anode might do the trick!

I guess my question is how do we keep Zout down with a choke load? Is a choke very specific to a particular valve? I've not played with choke loading and have no experience in the area, I obviously need to do some reading.

PS I found some more stuff on real worls TX so I will post something on the SUT thread when I get chance.

cheers,

-- Andrew
James D
Posts: 15
Joined: Thu Mar 31, 2005 2:00 pm
Location: UK

Post by James D »

Great thread :D

Two quick observations.

1) On where to place the load R with an mc step up.

I have been slapping my head as when I use an input trannie I always put the volume pot on the primary and unload the secondary... as it sounds better to my ears but I never thought to do it with an mc step up :oops: ... Well now I have and I like it 8)

2) I reckon the biggest benefit of the Aikido is its linearity with a low impedance output so, to my mind, it will greatly outperform a choke loaded triode as second stage in the RIAA - but try both ways round and see.

James
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi James,

I think you're possibly correct, but I wanted to offer something easy to build for those who don't want to go the whole hog with the Aikido, which is quite a challenge; a six tubed behemoth no less, with a planned PSU the size of a medium sized push-pull power amp, currently requiring 100mA. The heater circut for the Aikido variant needs 3 Amps, whereas 4 * D3A would need less than half that and be more than happy with a 70mA PSU.

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

Andrew wrote:Hi Dave,

I didn't explain myself very well, sorry.
actually it was me not paying close enough attention. I forgot that the termination was on the input of the LCR.
stage 2 needs to drive the line as well, I was think a choke loaded anode might do the trick!
what kind of line are you talking about? i'm assuming some cabling and a grid.
I guess my question is how do we keep Zout down with a choke load?
The Zout of a choke loaded stage is simply the value of the choke in parallel with the Rp of the tube it loads. Since in order to get full output at low frequencies the Z of the choke must be multiples of the tube Rp at the lowest frequency of interest so i just assume Zout=Tube Rp.
Is a choke very specific to a particular valve?
not really. the lower the Rp the better.

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hey Dave,

In my case the line-out is a TVC, but it could be an autoformer. I think ideally we're looking at something with a Zout of less that 2k. I tried several other phono's and found that most with a classic cathode follower output stage, say an ECC82, do a poor job with a TVC and can sound muddy.

My plan is to publish several designs for the phono to suit different systems and tastes, when I have built and evaluated them, of course. :D I know you're keen on choke loads and this is a new area for me to explore so I thought this could be a useful variant, provided the rest of the system fitted. My math seems to say that it doesn't require such a massive raised heater supply or as much HT current as the Aikido second stage.

Hey, just a thought, but with E810Fs throughout I reckon this could get close to the magic MC 60db without step-ups....now that might be worth trying...

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave,

Finally, started soldering it all together...I will most likely hack up a PSU for a meet with the guys in Early June before building a proper regulated supply.

There's nothing like a deadline to focus the mind.

Just an update on think, topology for the mk1 will be as follows....

Either D3A/E810F (sockets wired for both) running 26/30mA see what its like gain and sound wise into your LCR then small cap (starting with 100nF) in the Aikido (the one which was to be the first stage). Aikido is an ECC83 or 6072A. The 6072 is 6-7db down so I have to find it in the step ups or the front end to use this valve, which I would prefer as the ECC83 too expensive. Then into 6N30 or 6N6P Aikido Cathode Follower to really drive the TVC/autoformer.

That leaves the 6C45 Cascode for the front end and D3A triode back end, this could be choke loaded, but I will see how the Aikido drives the line first.

cheers,

-- Andrew
dave slagle
Posts: 2085
Joined: Sat Jan 22, 2005 3:54 am
Location: NYC
Contact:

Post by dave slagle »

hey andrew,

what cartridge are you running? I got to hear a bit of vinyl over the weekend and I have to admit that the one place i really like the concept of step up iron is with a Low Z cartridge.

I'm talking 1:50 or more with a 1-2 ohm cartridge and a minimal load to maximize gain.

refresh my memory as to what you are using for attenaution and maybe i can offer some more current ideas there too.

dave
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

Hi Dave,

Denon 103R (not so low Z - 14R, I think) or Dynavector Karat 17D2, better, I think.

I would certainly welcome the gain, if the Aikido idea takes off!

My problem is this, the Aikido has a gain of around 40% (roughly) of the mu of the first triode in the compound stage. So high mu is ideal, the second half of the stage is a line driver, so usual rules apply 6N6P, 6N30 and 5687 are all good here. The reason for Aikido is I'm using TVCs. I like these, but I'm also a firm believer in driving them with a low Z as possible to avoid treble loss. The Aikido gets down to 220R.

Interestingly, my early experiements have shown that current production ECC83 and are actually audibly little different in volume to NOS 6072A, which I would much prefer. The 83s also have poorer bass. However, theory says the 6072A should be 6db down over the 83s in this config and I suspect its the modern 83 that are coming down, if you see what I mean, to the level of the 6072A.

So....a few extra db gain from a step up might be really nice, but I also need to keep the Zout of the phono down. My calculations show a D3A into the Aikido (6072A into 6N30) with a 26db step ups nets about 56db total gain. I might get a bit more with E810F, that's what I'm hoping anyway.

But as you say, such a high winding ratio on a step up really needs a high drive/low Z cart, otherwise treble rolls off, even with primary loading.

My other thoughts are to get some anode chokes for D3As wound up, 200H to 300H at about 17-20mA might be about right, thus getting the Zout of the stage down that way, instead of using an Aikido.

I have almost completed the input stage in "bread board style" so if I get chance I post a photo later.

Next up is the RIAA networks....

cheers,

-- Andrew
Andrew
Posts: 201
Joined: Wed Apr 26, 2006 8:06 am

Post by Andrew »

OK as promised, progress photos..
Attachments
The first RIAA board with the smaller components on a daughter board.
The first RIAA board with the smaller components on a daughter board.
phono 001.jpg (246.31 KiB) Viewed 37895 times
Topside of the pentode inputs
Topside of the pentode inputs
phono 002.jpg (230.16 KiB) Viewed 37895 times
Backside of pentode input D3A or E810F catered for.
Backside of pentode input D3A or E810F catered for.
phono 004.jpg (279.54 KiB) Viewed 37896 times
Post Reply