I have a 250k load for the buffer stage after the RIAA amp. My calculated output impedance with a 22k anode resistor is about 11.3k. Dividing the 75us time constant by 11.3k gives a cap value of 6.6nF. A 6.8nF cap would be pretty close for me, but a little small for a 50k load. Still, it's a lot closer than 5.7nF.
Sheldon
That was not calculated, but measured on FFT with test disc, input transformer for MC, 6922, not 6DJ8, so don't take it as written in stone. I had good results up to 8200pF depending on cartridge and MC trafo. I think ECC88 like tubes have been reported as lower Rp 2K6-3K BTW.
Depends entirely on the operating conditions. I calculated from the plate curves, using the actual operating point. Even so, the cathode term dominates. The cathode resistance is Rk(u+1), or 562x34, for this stage (I mistakenly multiplied by 33 in my previous post, but the u can easily vary by that much). That comes to 19.1k. The plate resistance is added to that, as the plate is in series with the cathode. My calculation of 4.5k would give a total of 23.6k. If we take your lower estimate, the total resistance is 22.7k. So not that significant of a difference. The plate term matters much more if the cathode resistor is bypassed, as the cathode term drops out and the impedance for the tube is the plate resistance.
BTW, this is a good reason to use an unbypassed cathode resistor here. Not only is the distortion lower, but the plate resistance of the tube will drift with age. The u is much more stable. Since the u term dominates with an unbypassed cathode resistor, the impedance seen at the plate will be more constant.
However, if you want to lower the output impedance, you can bypass the cathode resistor. Then, calculate the required cap so that (anode resistor//plate resistance//next stage impedance)xRiaa cap = 75us. Let's assume a plate resistance of 4k at the chosen operating point, a 20k anode resistor, and a 50k load. Then 20k//4k//50k=3.42k. Then, 75us/3.42k=22nF. In this case, the output impedance is 20k//4k, or 3.33k.
Sheldon
Last edited:
Thanks both of you Salas and Sheldon for the interesting discussion of Zo for this preamp. I still have to digest much of what you have said by referring to some of the literature I have on the subject. From a practical viewpoint it seems I have the following options:
1. Increase the o/p coupling caps (C6-C7) to 1-2.2uF;
2. Bypass the cathode resistor of the last stage (R5-R21) with a 22nF cap;
3. Add a buffer amp.
Is this correct? With these changes, I'm aware each option has its own set of characteristics/problems. With opt 1 Salas says "1uF is kinda OK for 50K and cutting rumble on phono, still night and day VS 0.15uF to 50K". What do you mean here?
Option2 is said (by Sheldon) to increase distortion in the long term due to instability of Rp. We'll have to see how significant this will be.
A buffer means another project and adding another link in the chain. Any quick fix suggestion for a v-t or SS buffer? Sheldon I had a look at your version of the SB preamp, but your last stage (ECC88 srpp) doesn't look to me like a buffer, or am I missing/mixing something/s?
Thanks again for your great contributions on this forum (together with some others as well).
Cheers.
1. Increase the o/p coupling caps (C6-C7) to 1-2.2uF;
2. Bypass the cathode resistor of the last stage (R5-R21) with a 22nF cap;
3. Add a buffer amp.
Is this correct? With these changes, I'm aware each option has its own set of characteristics/problems. With opt 1 Salas says "1uF is kinda OK for 50K and cutting rumble on phono, still night and day VS 0.15uF to 50K". What do you mean here?
Option2 is said (by Sheldon) to increase distortion in the long term due to instability of Rp. We'll have to see how significant this will be.
A buffer means another project and adding another link in the chain. Any quick fix suggestion for a v-t or SS buffer? Sheldon I had a look at your version of the SB preamp, but your last stage (ECC88 srpp) doesn't look to me like a buffer, or am I missing/mixing something/s?
Thanks again for your great contributions on this forum (together with some others as well).
Cheers.
For option 2. Sheldon meant you change C1,C9 to 22nF if you bypass 562Rk with an electrolytic (I would use 470uF/16V), so the HF of the Riaa remains OK with new 3K3 Zo. But SB already gives enough THD as it is in his comments, that will rise, the gain will take another 12dB also, and I don't know about total noise. What about changing the Log pot in the C-J to 100K? If its possible, then no changes beyond higher couplers, just 1uF output capacitors will be plenty.
*My phrase meant that 1uF is much better than 0.15uF for 50K line input, and rather enough. Because it will cut enough infrasonic that turntables produce, maybe a good compromise.
*My phrase meant that 1uF is much better than 0.15uF for 50K line input, and rather enough. Because it will cut enough infrasonic that turntables produce, maybe a good compromise.
You have a fourth option, which I noted earlier (and which is the easiest to implement). That is, use a CCS load instead of an anode resistor for your final stage, and take the output from the top of the source resistor. http://pimmlabs.com/web/Active_loads_and_signal_current_control.html This will give you a low output impedance. If you can't change the pot on your pre amp, I would recommend this one. A fifth option, if your power supply has some current reserve, is to add a source follower stage as a buffer. I can draw these up later for you, with part values.
Option 2 will increase the distortion, because the unbypassed cathode resistor acts as a type of internal feedback to reduce distortion. Drift is a separate issue, and will cause your response curve to drift over time. If one channel drifts relative to the other, it will affect the imaging.
The final stage in my Bench phono amp is indeed a buffer stage.
Sheldon
Last edited:
Values for CCS: This is the easiest approach. Take a look at the depletion mode CCS cascode in the link above. You will adjust R to set the current. Measure the current across the anode resistor for the second stage of your phono amp to get this value. If you use DN2540, this will be about 400R for 5mA. You can use a trim pot here for set up, and replace it with a fixed resistor. If you do this, then just measure the voltage at the tube plate and adjust in situ. There are some refinements we can discuss if you choose the CCS.
No gate stopper is shown for the upper device, but you should use about a 1k resistor for this. BTW, you can use a small JFET for the lower device, and this will improve performance, as the capacitance will be lower. It will only see a few volts, so only the upper device needs to have a high voltage rating. The R will be different, according to the device chosen.
Values for Source Follower: See the first diagram in this link for the general schematic:http://www.geofex.com/Article_Folders/mosfet_folly/mosfetfolly.htm For this you can use something like the DN2540 (depletion mode device) or an enhancement mode Mosfet. I like the FQPF2N80 for it's low IDSS. You will want to run at least 5mA through this, and 10mA is better. Make sure your power supply can handle the extra load without a significant increase in ripple.
For this, the voltage at the gate will be the anode voltage of the Bench second stage, which is around 100V. Therefore the voltage at the source will be a few volts higher (for a depletion mode device) or a few volts lower (for enhancement mode). We will ignore that and call it 100 volts. So for 10mA you need a source resistor of 10k. Simple.
Sheldon
No gate stopper is shown for the upper device, but you should use about a 1k resistor for this. BTW, you can use a small JFET for the lower device, and this will improve performance, as the capacitance will be lower. It will only see a few volts, so only the upper device needs to have a high voltage rating. The R will be different, according to the device chosen.
Values for Source Follower: See the first diagram in this link for the general schematic:http://www.geofex.com/Article_Folders/mosfet_folly/mosfetfolly.htm For this you can use something like the DN2540 (depletion mode device) or an enhancement mode Mosfet. I like the FQPF2N80 for it's low IDSS. You will want to run at least 5mA through this, and 10mA is better. Make sure your power supply can handle the extra load without a significant increase in ripple.
For this, the voltage at the gate will be the anode voltage of the Bench second stage, which is around 100V. Therefore the voltage at the source will be a few volts higher (for a depletion mode device) or a few volts lower (for enhancement mode). We will ignore that and call it 100 volts. So for 10mA you need a source resistor of 10k. Simple.
Sheldon
CCS load for SB phono
Many thanks once again - you are most helpful!! I didn't realise the pitfalls of my option 2, especially with the possibility of inserting a CCS load instead of the plate resistor in the last stage - R3=20K. Didn't think about it because really I don't know how to do it. In fact I spent most of this time searching the forum for an appropriate ccs for this preamp but I'm lost to be honest.
Is it too much if ask you (real gentlemen!) to "cook" a basic, effective CCS for this riaa circuit?
As always THANK YOU.
Joe A
Many thanks once again - you are most helpful!! I didn't realise the pitfalls of my option 2, especially with the possibility of inserting a CCS load instead of the plate resistor in the last stage - R3=20K. Didn't think about it because really I don't know how to do it. In fact I spent most of this time searching the forum for an appropriate ccs for this preamp but I'm lost to be honest.
Is it too much if ask you (real gentlemen!) to "cook" a basic, effective CCS for this riaa circuit?
As always THANK YOU.
Joe A
Here is an example of a CCS, connected as a mu follower. The series output resistor sets the output impedance of the stage. The cap to ground is the RIAA cap, set to the resistor (1.6k x 47nF = 75usec). The 400R resistor should be adjusted to that the current matches what you have across the 20k plate resistor, which it replaces. You could use a good quality 1k pot here too. I left off the coupling cap after the CCS, but you will need on the output. The minimum value will depend on the input impedance for the following preamp, but 1uF will cover most any.
The second circuit is a source follower. It's simpler, but will draw extra current from your B+ supply. The output impedance is set by the series output resistor (220 Ohms here, but this is not critical). Everything else stays the same before it, except the coupling cap, which is after the follower.. I would change the final RIAA resistor to about 7nF, instead of 5.6nF.
You can use IXYS 10M45 in the CCS, but the source resistor will need to be adjusted. Check the data sheet for an initial value.
You can use the IXYS or DN2540 for the source follower too, with no value changes. Other Mosfets than the one I show can also be used, but pick one with similar low Crss (reverse transfer capacitance).
Salas can pitch in with any comments too, as he knows circuit design better than I do.
Sheldon
The second circuit is a source follower. It's simpler, but will draw extra current from your B+ supply. The output impedance is set by the series output resistor (220 Ohms here, but this is not critical). Everything else stays the same before it, except the coupling cap, which is after the follower.. I would change the final RIAA resistor to about 7nF, instead of 5.6nF.
You can use IXYS 10M45 in the CCS, but the source resistor will need to be adjusted. Check the data sheet for an initial value.
You can use the IXYS or DN2540 for the source follower too, with no value changes. Other Mosfets than the one I show can also be used, but pick one with similar low Crss (reverse transfer capacitance).
Salas can pitch in with any comments too, as he knows circuit design better than I do.
Sheldon
Attachments
Last edited:
Just to make sure I have understood you correctly, we have two options to achieve a similar result, i.e. to lower Zout - a CCS to replace the anode resistor or a Source Follower which basically adds a buffer stage to the SB preamp. From my viewpoint the SF is the simpler circuit to add on to the preamp circuit.
I think I will use the DN2540 which is more readily available. In that case does the 10K source resistor still apply?
On the implementation side, I will break the C1-C6 connection in SB's circuit, then connect the gate to C1 and connect C6 to the source. Is this correct?
Do you still think I should increase C6 from 0.15uF to say 1uF?
As regards my power supply I'm using Salas' shunt regulator and I believe the transformer is up to the task.
Thanks for your patience with my newbie questions.
Regards,
Joe A
You've got it.
Yes.
Yes.
This depends on the input impedance of the following preamp. You'd like the roll off to occur below 10Hz. For 50k impedance, a 0.47u cap gives a -3dB point at about 7Hz. A 1uF cap would be fine here.
Sounds good. Check with Salas on adjusting the regulator current to account for the added load. As long as your transformer is up to it, and you can handle any extra heat, then this is the easiest solution. BTW, you don't have to take the power supply B+ for the source follower. You can take it before your regulator, because the follower has very good psrr.
Sheldon
Last edited:
Excellent. Many thanks, Sheldon. Feel more confident now. Will report after I make the modification.
Regards.
Regards.
Better to post here. Working through things helps others, and allows for better answers too.
Sheldon
Sheldon
OK Sheldon, Thanks. Point taken. So far I took my SB Phono to insert a Source Follower after the last ecc88 stage as per post #230. Fried two sets of source resistors and possibly also the DN2540s. At Sheldon's suggestion I removed the SF and made voltage checks around the preamp and got some consistently confusing readings as follows:
Supply voltage is 185V from one Salas HV Shunt Regulator. OK
6N23P plate: pin 6: circuit voltage 102; my readings 69 & 70 volts
6N23P plate - pin 1: circuit V = 123V; reading - 32V & 32V
6N23P grid - pin 2 : circuit V = 85V; reading - 17V / 17V
6N23P pin 8: circuit 2.5V; reading - 1.8V / 1.8V
6N23P pin 3: circuit 87V; reading - 39V /39V
12AY7 pins 8/3: circuit 1.91V; reading 1.5V /1,5V.
I changed the valves in case I might have damaged them when I installed the SFs. First the 6N23P were changed with ecc88 and the same readings were obtained. I then also changed the 12AY7 and the readings remained basically the same.
What could have gone wrong? Before I installed the SFs I had a working preamp, though I didn't know what the voltages were then. Obviously I haven't run the preamp in my system with these (lower) voltages.
Thanks in advance for any help you can give me.
Joe A
Thought about the possibility of some loose ground and checked. Everything looks OK. Then I checked the filament voltage which runs on a 317 regulator in constant current mode as described earlier by Salas. Here the voltage was a bit less than 5V (4.8 maybe).
I haven't yet fixed this but would like to ask whether a low filament voltage could result in the low voltage readings around the valves reported in the previous post? At the same time I wonder why it drifted to just 5V when it was originally 6V and over.
I haven't yet fixed this but would like to ask whether a low filament voltage could result in the low voltage readings around the valves reported in the previous post? At the same time I wonder why it drifted to just 5V when it was originally 6V and over.
Sheldon at #230 showed an enhancement type Mosfet for source follower. You used a Depletion? Try an IRF820 if not having the low Crss one suggested. Its near. Or any IRF NMOS with enough voltage spec just for testing.
Those numbers are indeed confusing. Let's see if we can sort this out. First thing to do as a check is measure all your resistors in circuit, and measure every node to ground. Do this with the schematic in front of you. Make sure you don't make simple easy to miss errors, like the wrong number of zeros on the meter (been there before). Also remove all the tubes and make the same measurements as above, with the supply on. As to your numbers:
If you supply voltage is 185 and your plate voltage is 32V for the first section, and the plate resistor is 100k, then the current through that plate is 1.53mA. Therefore you should have 1.53mA through the cathode resistors on the AY7, as they are in series. The total cathode resistance is 3.2k. Therefore you should see a voltage drop of 4.9V, or a measurement of 4.96V at the cathode. You report 1.5V, so one of the measurements is wrong, the resistors are wrong, or that current is going elsewhere. I don't think it likely that your cathode cap or RIAA cap have failed, as it's the same on both channels. An important point, BTW, measure voltage, but think in terms of current. Current flow must always add up.
On the AY7 grid, you report a reading of 17V. It should be 85V. No current should flow through the grid. This voltage is divided down from your B+, so the divider is wrong, or current is flowing elsewhere. Again, it could be the decoupling cap here, but unlikely for both channels to fail, unless both are done wrong in the same way.
You show pin 6 as 70V. This implies a voltage drop across the 20k resistor of 115V. The current is therefore 115V/20k = 5.5mA. This current should all flow through the 562R cathode resistor. So, 0.0055x562=3.1V. You measure 1.8V. Again, wrong resistor values, wrong measurement, or current flowing where it shouldn't. Are you sure your meter is operating properly. Need new battery?
The low filament voltage will not cause the problems you see. It's not low enough to affect your circuit measurements. But it should not have changed that much.
Hi Salas,
Yes, I showed an enhancement part, but depletion is fine too. It will shift the voltage at the source, but it's only a swing of +/- 4V or so. I recommended in PM that Joe use a protection diode, between gate and source. The diode orientation must be switched to account for the type of device, but otherwise the operation is the same.
Final point Joe,
Tubes will tolerate much more abuse than SS devices. For SS, you must respect the specification limits. If you exceed them for just an instant, they will fail. If you are not sure of your skills, buy some cheap parts and plenty of extra. Do one channel at a time. When you've got things working well, you can plug in better parts if you need them.
Sheldon
Got pictures of the wiring?
Last edited:
- Status
- Not open for further replies.