|
Audio Asylum Thread Printer Get a view of an entire thread on one page |
For Sale Ads |
Inmates,I think I sort of know what the answer to my question is, but want to get some reassurance from the group. I have done a lot of research on both composite and split RIAA networks and have settled on a composite network based on my personal preferences and prejudices ;0) . I have read everything I could get my hands on (i.e. Morgan Jones, Allen Wright, Simon Shilton, et al) but did not get a sense of the impact the first two resistors (R1 & R0) make on sound. Please see below link of the circuit for clarification. The trade off I am wrestling with is gain versus grid leak resistance of following tube. If I make R1 larger relative to R0, I will get less gain because R1 & R0 act as a voltage divider. However, the grid leak resistance for the following tube will be good. I always try to keep grid leak resistances to a minimum. On the other hand, if I make R1 and R0 equal to each other, but twice that of R1’, gain will be better, but R0 will more than likely be over 500K. The tube that will be driving the RIAA network is a constant current source loaded 6GK5 with Vp=160V, Ip=16mA, and Vg= -1V. At these operating points the 6GK5’s Gm=16.8mA/V, Mu=80, and Rp=4.8K. Therefore, as you can see it will not have any problems driving the RIAA network, but I’m worried about grid leak resistance of the following stage. So, what do you Inmates feel like would do more damage, sacrificing gain that may necessitate another gain stage, or try living with a very high grid leak resistance? All your comments are welcomed and appreciated.
JLH
http://www.users.globalnet.co.uk/~valveamp/RIAAStageDesign.htm
Follow Ups:
What I've done is to first choose the source impedance for the RIAA network (plate resistance of the first triode) and the grid ground return for the second stage. Then the RIAA can split the difference, so to speak.Note that the source resistance adds to R1, while the grid return R0 is in parallel and efectively reduces R1. With the right combination, the net effect is zero, so the actual value of R1 is the same as if rp=0 and R0=infinity.
In practice, there are only a few combinations of available resistors and capacitors that work out to a pretty good RIAA without fussing. Given one of those (close to the target found above), and rp of the first stage, you can calculate an R0 for the second stage.
Despite my daft idea for using the large coupling cap before the RIAA network, I think it works fantastic. See the schematic below for my phono stage design, then replace things as follows:* Change bipolar CCS with new MOSFET CCS in http://www.sonic.net/~ktstrain/MOSCCS.jpg
* Change V1 to 6ER5, also change its cathode resistor where you want the plate voltage and current to be (I chose 90V and 2.5mA for Vp and Ip) and remove the cathode bypass cap entirely.
* Change V2 to 6GK5, change the cathode resistor again for your desired operating point (I chose 110V and 2.5mA) and remove the cathode bypass again.
* Change the first coupling cap to two 0.22uF TRT teflon caps in parallel.
* Change the 0.01uF cap in the RIAA to a TRT teflon.
* Change the 0.22uF output cap to TRT teflon.
* Change all resistors in the RIAA to Caddock TF020 film. Change cathode resistors to Caddock MK132.
This is a MM input phono preamp. Use a high quality MC step-up transformer for MC input. I never heard tube or transistor do as good a job as my EAR MC-3 for this function - zero noise, great sound.
That's what I use. I think it's totally first rate. Better than anything else I have heard. But's it's an expensive setup with those parts.Kurt
Hi Kurt,> Despite my daft idea for using the large coupling cap before the
> RIAA network,I guess I must explain the "daft idea" statement.
If you place the coupling Cap first you remove the bias from the RIAA EQ Capacitors. The results are invariably not only a much greater sensitivity to capacitor quality, but even with polystyrene/Teflon Cap's a loss of transparency that I find difficult to accept, given the minimal price to pay for doing it the other way.
The interactivity between coupling cap and the lowest RIAA EQ timeconstant I personally consider as distant second in the race, but both effects (in their respective sonic areas) are distinctly audible.
As someone who has tried it both ways I know which way my cookie crumbles on that.
Later T
That's a valid concern I didn't consider. I will think about this some more and maybe try something that biases up the caps.Kurt
I rearranged everything. There's now one 0.22 uF cap between the second tube and its grid resistor (I am using a high 2M resistance there) and the caps are all charged up by the plate of the first tube now. I simulated that there was a small 0.2 dB difference drop at 40 Hz by moving this over, but I had to use a large 2M grid resistor to keep it that low. It works.It sounds a little different. Some of the dynamics seem diminished right now, and I think this is a breaking in period for the caps finally. They've never seen this voltage before. I'll keep playing it like this.
Kurt
Hi,It is possible that the what you hear is breakin, alternatively, 2M strikes me as quite high in Value for the gridleak, you may get a little "gridleak bias" now, shifting the operating point (the CCS load makes the Valve more vulnerable to this....
If you shifted the operating point correct the cathode resistor for identical anode voltages....
Ciao T
You're right. They do have some grid leak biasing going on. I have a small trimmer resistor built in there now for adjusting the bias voltage. Not the finest sounding cathode resistor, but it works well to get a stable bias. I adjusted it for 140V on the plate and it stays put after warm-up.Kurt
Hi,
Should I copy your design?
Thinking of putting a riaa-amp on the underside of the armboard on my 124.
A two tube 10mA in total amp would fit the bill.
[If I ever manages to make up my mind :-) ]
Is it worth the hazzle?mvh /Pär
> > Should I copy your design? < <I don't know. I'm still tweaking on it. When it gets ready for another revision release I'll draw up a new schematic. Things are changing right now. You need to know a few things about implementing low noise circuits to be successful with this.
Kurt
Hi,
Yea... Low noice is a drag (to me).
I will follow your postings on this riaa in the future (though I'm a bit biased towards a three stage riaa but in this application heat and space (not psu) is a bit limited so...).mvh (Pär
Hi Kurt,You realise of course what that means. Just bump up the Gridleak resistor up to around 4M7...10M (works well for me) and eliminate the Cathode Resistor/Capacitor combo and use a much smaller value (and thus higher quality coupling capacitor....
But I think first you should make sure you really like the RIAA EQ this way around, then you can move on.
I know loads of people frown upon gridleak biasing, but it is not such a bad idea actually....
Ciao T
Kurt,How about the Vishay bulk metal foil trimmers that Percy stocks? Maybe a series combination of a Riken or Kiwame audiophile Carbon plus the Vishay for adjustment would sound good.
Eli D.
> > * Change bipolar CCS with new MOSFET CCS in http://www.sonic.net/~ktstrain/MOSCCS.jpg < <Sorry, I haven't made this change yet, but I suspect it to be better. I am not fully sure about the noise level from the FETs being low enough.
I'm designing a RIAA preamp myself. I've decided that getting the gain up(no pun intended)to a workable level before I monkey around with the EQ is my best road to venture down. I'm using a cascode. I know they produce more even ordered harmonics but I would rather listen to that than odd ordered harmonics. I'm using an unequal cascode with a hi-mu-high transconductance tube on bottom and a 12AX7 for the grounded grid. This is DC coupled to a second stage common cathode-anode follower which is DC coupled to a cathode follower. There is only one capacitor in the feedback loop and I may get rid of that with a servo amp. Thought I would split up the RIAA part feedback and part passive. The cascode produces enough gain that I can drive just about any passive filter and have plenty left over. PressON!Soldier of fortune James
Hi,I have been around the same bends a few times. The BEST solution is split RIAA in a three stage design, which removes the need for compromises, especially if you use high Gm, low Rp Valves. I would probably consider RLC (Tango style) as better, but have so far not heard enough using this principle to make solid call.
Another option, especially if you place the coupling capacitor BEFORE the RIAA Network (which is IMHO a particulary daft idea, but never mind) is to have the Gidleak resistor there and then to follow with the RIAA EQ, all interactivity is removed, other than that the gridleak resistor plus the RIAA EQ series resistor must be lower than the maximumpermissable grid resistance.
Or perhaps not - if you use a vary large Value Grid Resistor you can use fairly low value coupling Cap's AND dispose with the cathode RC combo, by gridleak biasing the Valve....
Ciao T
It's been more than 40 years since I messed with contact (grid leak) bias. As a youngster, the tube circuits I built (AM radio) invariably used contact bias. The grid leak resistor was always 2.2 MOhms, even on a 6AQ5 "final" which had supplementary cathode bias. Obviously, my experience was not HIFI, but my memories of a 6AN8 using the pentode as a grid leak detector and the triode as a driver for the 6AQ5 are clear. The thing worked fine with no cathode resistor on either section of the 6AN8. IMO, it's definitely worth a try.
Eli D.
My Counterpoint 7.1 uses grid leak bias in the phono section, so I've been trying to find info on it's use. So far, I've only found material from the "NEETS" course, which concludes with -"The main limiting factor(of grid leak bias) is the amount of distortion that you may be willing to tolerate. Distortion occurs during the positive alternation when the grid draws current. Current drawn from the electron stream by the grid never reaches the plate; therefore the negative-going output is not a faithful reproduction of the input, while the positive-going output (during the negative input cycle) will be a faithful reproduction of the input."
The remarks about distortion are correct when signal levels are substantial. The charge built up on the grid has to be large enough to stay negative throughout the entire voltage cycle for operation to be "linear". An AM envelope detector and phono stages are situations where the LOW signal level allows contact biased tubes to work well.If the grid leak resistor was omitted, the tube would cut itself off quite quickly, as more and more electrons collide with the grid and the charge builds up. The idea is to establish a balance between electrons striking the grid and those going to ground via the grid leak resistor. 2.2 MOhms is the "classic" value in AM envelope detectors. It works well with all sorts of tubes, both triodes and pentodes. The value of the grid leak resistor that works "best" in any situation would have to be determined on the bench, but 2.2 MOhms is a good place to start.
Eli D.
This post is made possible by the generous support of people like you and our sponsors: