| etalon90 |
HI all
I just finish my LCR riaa stage.
I love it but it`s not still perfect.
I need a better bass, more robust and extended...
that is the only complain i have.
R9,R12,R13 and R11 are inductor DCR
On simulation, I have 1.2 Db riaa accuracy. I just don`t understand why my bass is not as good as my previous riaa project.
PS; I use sealed acid battery for PSU (B+ and heater) I expect my psu to have about 24ohm impedance (12battery in serie, about 2 ohm each)
Thanks |
|
|
| analog_sa |
Wow, that's what i call impressive!
Are the numbers next to the chokes Henry? 30H may be a bit low for a 417. What types of chokes are you using? They must be pretty well shielded if you don't complain of hum.
Bass? Some people may find 1.2db excessive. What did you use to simulate and calculate the riaa values? Have you tried adding capacitance and chokes in the PS?
I wish i had the guts to build one like that! |
|
|
| ashok |
| quote: | | I wish i had the guts to build one like that |
Me too ! |
|
|
| etalon90 |
>Wow, that's what i call impressive!
Thanks,
>Are the numbers next to the chokes Henry? 30H may be a bit >low for a 417.
I run at 16mA. So -3db is 17hz. I don`t use bigger choke because i`m affraid to heve subsonic noise.
>What types of chokes are you using? They must be pretty well >shielded if you don't complain of hum.
With battery supply, i don`t have ANY hum!!! ;)
>Bass? Some people may find 1.2db excessive. What did you use >to simulate and calculate the riaa values?
Orcad capture lite (free on their official website)
Have you tried adding capacitance and chokes in the PS?
Should I?. If I add capacitance, that will be electrolytic one. I tend to don`t use them...
I wish i had the guts to build one like that!
I greatly recommend to build a LRC type. that better to passive RC EQ. I changed the value of mine for fit the output impedance of my first stage (1.8K)
this is the original value from tango EQ.
martin |
|
|
| EC8010 |
Moving magnet cartridges expect to see 47k, rather than 100k. It won't affect the bass, but it will cause a rise in treble towards 20kHz that could make the bass seem lightweight in comparison.
2 Ohms for a lead acid battery seems very high. Are you sure? |
|
|
| fdegrove |
Hi,
Could it be you lose bass response because the LCR RIAA isn't terminated?
Cheers,;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
Could it be you lose bass response because the LCR RIAA isn't terminated? |
I wondered about that, but etalon90 says he ran a simulation on it. Mind you. 1.2dB error is rather large... |
|
|
| fdegrove |
Hi,
| quote: | | Mind you. 1.2dB error is rather large... |
Exactly.
I know that proper termination of the 600 Ohm ones is mission critical; put 620 Ohm instead of 600 Ohm to terminate it and you lose the entire RIAA curve along with it.
I see no reason why this wouldn't be the case with a 1K8 cell.
Either way it doesn't take much to find out and if it works fine this would be great news.
After all a 1K8 Ohm load is much easier to drive with valves than a measle 600R, right?
Cheers,;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
After all a 1k8 Ohm load is much easier to drive with valves than a measly 600R, right? |
This is a common misconception (I've even found it on semiconductor manufacturer's data sheets). 600 Ohm meant driving from 600 Ohm, and terminating in 600 Ohm. In other words, the source actually saw a load of 1200 Ohm. (The source was always made to have almost zero output resistance, so that adding a series 600 Ohm resistor gave constant 600 Ohm output resistance with frequency and level.) |
|
|
| fdegrove |
Hi,
| quote: | | (The source was always made to have almost zero output resistance, so that adding a series 600 Ohm resistor gave constant 600 Ohm output resistance with frequency and level.) |
O.K. I see the mistake.
For this particular implementation (LCR RIAA) g' would than be 3K6 which is still a very heavy load for anything but a xformer or CF to tackle, correct?
Which is why we almost invariably see a stepdown xformer after the driving stage in pro gear a la Pultech etc.
Any advice on how to best implement this LCR correction without resorting to esoteric xformers?
TIA,;) |
|
|
| EC8010 |
| quote: | Originally posted by fdegrove
Any advice on how to best implement this LCR correction without resorting to esoteric xformers? |
I'm afraid not. You need a low working impedance, otherwise the inductors become unreasonably large. It really is the ideal place for a transformer - unless (just thought of it) you were to use an OTL headphone amplifier, perhaps with E55L or 6C45? |
|
|
| SY |
| EC, perhaps some sort of gyrator circuit might do instead of the iron? |
|
|
| EC8010 |
| Perhaps. But it a gyrator would inevitably add noise which couldn't really be tolerated in the first stage of this design. If you added more gain before the equalisation, you could get away with it, but then you would have lost the appealing simplicity of the original design. |
|
|
| SY |
| OK, fair enough. But since RIAA can be implemented simply and inexpensively with R and C, I guess I don't see the point of having inductances in there except for some esthetic ideal. Of course, I could say the same thing about the tubes, but I'd be run out of the forum on a rail. |
|
|
| fdegrove |
Hi,
| quote: | | Of course, I could say the same thing about the tubes, but I'd be run out of the forum on a rail. |
On the bright side, you could always buy ms Y a girator for her next birthday.
I hear they're excellent for making mayonaise...;) |
|
|
| SY |
Mrs Y doesn't cook. Dr Y does. But never mayonnaise, and never on frites; that would be uncivilized.
So, Frank, you'd be the logical one to ask: assuming decent first stage gain to get the signal out of the muck, what is the advantage of LC versus RC equalization? |
|
|
| fdegrove |
Hi,
| quote: | | what is the advantage of LC versus RC equalization? |
Technically I don't see any.
From people that built a preamp using such a LCR correction I hear nothing but praise.
Comments like it sounds like that good old Decca without the flaws are not uncommon.
| quote: | | Mrs Y doesn't cook. Dr Y does. |
Making mayo isn't easy without the right utensils...
Cheers or should that be bon app... |
|
|
| EC8010 |
| quote: | Originally posted by SY
OK, fair enough. But since RIAA can be implemented simply and inexpensively with R and C, I guess I don't see the point of having inductances in there except for some esthetic ideal. |
I'll probably be run out of the forum on the same rail as you, but in a bridged T equaliser like this, the equalisation is actually being done by the CR networks, and the LR department is the complex conjugate that forces the impedance to be resistive. Or should be. To do it properly means matched impedances at either end.
Now, a 3k6 resistive load is a nasty load, and will cause 2nd harmonic distortion to rise, but it isn't as nasty as a reactive load that forces the loadline into an ellipse. That really causes distortion. So, yes, there could be a demonstrable technical advantage to LCR equalisation if implemented conscientiously. |
|
|
| SY |
That's interesting, and you've now earned more questions by being sucker enough to answer me.
Since in a typical tube phono stage we're swinging some really low voltages compared to what's dropped across the plate load and the tube (maybe a couple hundred millivolts compared to maybe a couple of hundred volts), how much distortion, more or less, is really coming from the tube pumping into the reactive load of a passive RC RIAA network? How much reduction, more or less, could be available by putting in these conjugate networks?
My first stage tube doesn't see anything like 3-4K from the following passive RIAA, it's something like 20-30K at minimum, and I don't think I've done anything out of the ordinary. If you think there might be an advantage to get the reactances out of the plate circuit, I could always direct couple the first stage to a cathode follower- cheaper and easier than all those inductors, and I don't have to lay awake nights worrying about using magnetic materials in the signal path. |
|
|
| EC8010 |
| quote: | Originally posted by SY
That's interesting, and you've now earned more questions by being sucker enough to answer me. |
Sigh. Rod for my own back...
| quote: | | Since in a typical tube phono stage we're swinging some really low voltages compared to what's dropped across the plate load and the tube (maybe a couple hundred millivolts compared to maybe a couple of hundred volts), how much distortion, more or less, is really coming from the tube pumping into the reactive load of a passive RC RIAA network? |
There are whacking great signal voltages leaving the stage before equalisation! Remember, RIAA equalisation has a dynamic range of 39dB from 20Hz to 20kHz - and that's without considering the dynamic range of the recorded signal. RIAA pre-amplifiers are as challenging as microphone pre-amplifiers. RIAA design is a juggling act to try to encompass the dynamic range of LP (say, 60dB) plus the dynamic range of RIAA (39dB) = 100dB, without running into noise or headroom problems. It's why valves are ideal.
| quote: | | How much reduction, more or less, could be available by putting in these conjugate networks? |
Very hard to say without designing an entire pre-amplifier with them in mind, but it might be worth it...
| quote: | | My first stage tube doesn't see anything like 3-4K from the following passive RIAA, it's something like 20-30K at minimum, and I don't think I've done anything out of the ordinary. |
Yes, that's fairly typical.
| quote: | | If you think there might be an advantage to get the reactances out of the plate circuit, I could always direct couple the first stage to a cathode follower- cheaper and easier than all those inductors, and I don't have to lay awake nights worrying about using magnetic materials in the signal path. |
Think current, rather than voltage. The cathode follower still has to swing current into the unpleasant load. It has lots of local negative feedback (am I allowed to approve of such a thing?), but it still has vertical excursions on its curves. Nevertheless, yes, a high current (perhaps 10mA) cathode follower can make sense for buffering an RIAA equalisation network - especially if the preceding stage has been optimised for linearity and uses an active load or weedy (but low distortion) valve such as 7F7. |
|
|
| SY |
Oh, and you thought that being helpful would be enough for me to stop pestering you. Naive.
| quote: | | There are whacking great signal voltages leaving the stage before equalisation! |
How great is "whacking great"? A typical mm will have an output of roughly a millivolt per cm/sec. Max velocities before mistracking are on the order of 30 cm/sec for the better examples, and those velocities are the exception. So I see a max input of something like 30 mv. With a triode in the first hole, we will expect something on the order of a gain of 40-50, a bit less for the triodes I use, a bit more for tubes like 12AX7. So that looks like a max swing at the plate of something like 3V or less.
Now, one might argue that you can see higher outputs when a cartridge mistracks. But the distortion of the phono preamp becomes a pretty minor issue with that diamond chattering about, as long as the designer has thought about overload recovery or put in a bit of bandwidth limiting to get rid of the ultrasonics.
What have I missed?
| quote: | | The cathode follower still has to swing current into the unpleasant load. |
Well, it's not that unpleasant. Do you really need 10 ma? With a 20K minimum load, that corresponds to 200V, which, I grant you, IS a "whacking great" amount. |
|
|
| EC8010 |
It's not that you need the full 10mA for audio swing, it's that vertical swings are so non-linear that they have to be made a small proportion of the quiescent current. In order that the noise of later stages doesn't significantly add to that of the first stage, you need quite a healthy level before applying your equalisation (which must introduce a loss of 19dB).
There used to be a rather nice graph of recorded velocities (Shure?), which showed double the velocity you mention. But the killer is dust and scratches which mistrack to produce even higher amplitude signals. If the pre-amplifier overloads on these, it prolongs the effect. To my ears, the main difference between RIAA pre-amplifiers is in the relative level of clicks and scratches. |
|
|
| SY |
| Ah, thanks. The rails are getting a bit smoother now, we must be out in a rural area. My idee fixee for a long time has been that any demonstrable audible differences between competently designed amps/preamps/boxes of gain is overload recovery. |
|
|
| EC8010 |
| quote: | Originally posted by SY
My idee fixee for a long time has been that any demonstrable audible differences between competently designed amps/preamps/boxes of gain is overload recovery. |
I think you're right. If you look at the number of people using pre-amplifiers with gain between CD players and power amplifiers, you have to conclude that they drive their power amplifiers into overload much of the time. The least tolerant of overload is traditional silicon, followed by valve push-pull with feedback, valve push-pull without feedback, single ended valve. |
|
|
| etalon90 |
Sorry for delay in answering.
My computer was down and I format my 2 hard disk...
I still don`t know ahy my bass il weak (not as mutch as you can think)
I exclude plate choke because i already use them in my dac. The bass in my dac is ok.
I do 2 error in my schematic. The Zin is not 100K butk 50K (i use two 100k in parallel) And the running current is not 16mA burt 26 mA.
I include a frequency reponse of the value I simulated. |
|
|
| etalon90 |
Now. I try to add a 1.8K after the LCR for adding load as suggested by a member. (don`t remember witch one...)
All Eq became worse. Around 8Db error on curve.
I remember when I designed this RIAA setup I had 2 option.
1-A 0 impedance source driving the LRC(600ohm) and finish by a 600 ohm load. then, next stage. This is why we see a cathode folower driving a stepdown transformer to keep Zout as low as possible.
2- a out of 1.8K (of the Zout of your driver, 1.8K in my case) This drive a 1.8K LCR without a ending load. As you see in my other post, the theorical value are good.
I run a simulation and probe the current in the capacitor (6.8uf of the first stage) . As you see, it draw less current in bass area than in high frequency. So the weak bass don`t came by dre driver saturation or difficulty to drive LCR.
I run a test. I add a serial resistor on my powersupply to see the effect on sound. My battery+a 50ohm in serie. I didn`t notice any change in tonal balance. So adding big capacitor on the B+ should not help the bass.
I`m not an expert but I thinkink remove the 470Uf on the cathode and running the 2 stage in superpath. What will be the effect on local feedback (I don`t know how to calculate local feedback in tube circuit.)
I have another theorie for my weak bass. The value of the inductor are nominal value at nominal current. So a LCR don`t draw nominal current in choke. In this case, the value of my inductor will not be correct. My simulation program don`t take this phemomen in account. That could be my problem... but how to fix that...
Does it make sense? |
|
|
| fdegrove |
Hi,
| quote: | | Does it make sense? |
Errrr...Not to me but then I never really understood how you got it so close to RIAA spec in the first place.
From what I've seen done with the 600 Ohm version feeding a sim with inverse RIAA network it can be tweeked to ruler flat response.
Every single person writing about these LCR seem to be very insistant on the importance of proper termination.
So I am left very puzzled indeed.
Cheers et bon courage,;) |
|
|
| Mike C |
| quote: | Originally posted by SY
How great is "whacking great"? A typical mm will have an output of roughly a millivolt per cm/sec. Max velocities before mistracking are on the order of 30 cm/sec for the better examples, and those velocities are the exception. So I see a max input of something like 30 mv.
What have I missed?
|
What you've missed is the RIAA pre emphasis.
We are always thinking of RIAA in terms of equalisation, in terms of
the severe HF roll off that we apply.
But don't forget that an equal HF boost will have been applied to the
signal before cutting the record.
A typical MM cartridge will have a nominal output of about 5mV, quoted
at 1kHz.
But a similar signal level at 20kHz will have been boosted by almost
20dB. What comes off the record at high frequencies can be 20dB
higher than 'nominal'.
And that's why you're about a factor of 10 out ....
You need a further correction. The cartridge figures above are RMS
(well, that's how they are usually quoted).
The 30mV you worked with has already become best part of 300mV RMS;
and that's roughly +/- 420mV peak. When looking into overloads, we
are interested in peak values rather than RMS.
For a touch more headroom, +/- 0.5V looks like a good design spec to me.
In my case I bias the 1st valve at -1.0V so as to allow 0.5V swing
comfortably without grid current.
This represents +/- 20V going into the RIAA stage; in the nicest way,
is that 'whacking great' enough for you? :)
Thanks are due to Par on Tube DIY Asylum; his comments to my early
thoughts on Isabelle made me look into this a bit further!
It's nice when we can all help each other. |
|
|
| SY |
The RIAA pre-emphasis is, I think, a bit of a red herring. Magnetic cartridges respond to stylus velocity, irrespective of what the following electronics do. The max velocity that normal cartridges can track, even at high frequencies, is still about 30cm/sec (I got that number from several Audio reviews of cartridges and some material Shure included with my test disc). So that corresponds to 30 mV, whether at 1KHz or 20 kHz.
Going from RMS to peak is a 41% change, I'll grant you, so my 30 mV figure could well be 42 mV peak. EC8010 emailed me a reference that studied actual outputs using a cumulative envelope detector- I'll be interested in reading this. On my own, I looked at max swing at the plate of the first tube in my RIAA stage (no feedback loop to confuse things, stage gain is 110) using a scope with some tracking test records as the source. The most I could see was a 3V swing, and that was exceptional, occuring as my poor cartridge tried to negociate the highest level test bands on the Shure disc. |
|
|
|
