With cathode bias the bias voltage is generated at the cathode. The grid stays at (or very near) ground potential. Everything is relative to the cathode. If the tube 'sees' the cathode at fe. +10V,while the grid is at ground,it thinks the grid is at -10V. (remember,relative to the cathode.) As far as the tube knows,the cathode is at ground,and the grid is at -10V,but it's not. We've tricked the tube into thinking otherwise.
Clear as mud?
Some mention has been made of Blocking and how grid chokes help recovery from periodic A2 (grid current) operation.
It is my contention that grid chokes actually make the recovery slower and introduce substantial LF ringing (at the CL resonant frequency) from periodic A2 overload. I know this flies in the face of 99% of what has been said about grid chokes and A2 overload but that is my story and I'm sticking to it.
I do want to be clear that I am not making any judgments on the perceived sonic qualities of grid chokes and in fact I like them particularly when tapped for attenuation The only thing I am trying to strike up debate on is the whole A2 recovery issue.
dave
It is my contention that grid chokes actually make the recovery slower and introduce substantial LF ringing (at the CL resonant frequency) from periodic A2 overload. I know this flies in the face of 99% of what has been said about grid chokes and A2 overload but that is my story and I'm sticking to it.
I do want to be clear that I am not making any judgments on the perceived sonic qualities of grid chokes and in fact I like them particularly when tapped for attenuation
The only thing I am trying to strike up debate on is the whole A2 recovery issue.dave
Did you ever resolve that whole debate with Mike LaFevre over whether the low frequency peak predicted in Spice really occurs with very large grid chokes? I recall doing some measurements with the 150H Hammonds suggesting it's nowhere near as severe as predicted but long ago lost the data.
Interesting. As a beginner I find CCS much simpler than any other load. You reduce the number of unknown, at least you know what current you get from it. When combined to diodes are the best. You fix another parameter. Not really sure how you calculate the impedance of a CCS.
Thanks for the interesting info,
D.
Thanks for the interesting info,
D.
Whether or not you get the predicted peak depends on a number of things, the most important is how accurately you modeled the circuit:
- Grid choke inductance and dcr
- How accurately the source impedance of the driver is modeled
- If you are interested in HF behavior you need to know the effective shunt capacitance the choke presents to ground.
Knowing these things I have found there to be an excellent correlation between the predicted and measured response - and this often results in the need to use very large coupling caps to tame the subsonic resonance if the choke Q is high (dcr low) which then creates a problem with blocking distortion if you attempt class A2 operation.
I suspect the hammond chokes have relatively low Q and if also driven by a relatively high source impedance may not exhibit a very large subsonic peak with undersized coupling capacitors. (The Q of this resonant circuit is strongly influenced by the choke dcr and the source impedance driving the CL coupling circuit.) I am also pretty confident that this can be modeled accurately if the right parameters are known.
I am currently working on some of the same issues Dave has mentioned and have concluded that I will be using ITs for the reasons mentioned. (My first application.)
Last edited:
Signal ground with respect to system safety ground. So if you have noise on your system safety ground, the signal ground will ride on that noise and present it to the signal of the following stage.
Picture taken from Dave Davenports paper on grounding
Attachments
Last edited:
Bas that picture is very unclear. It looks as if the signal ground of both stages is referred to the 'thick' line, and in that case there is no noise injection - the noise is common mode. However, if the receiver amp has its reference ground connected to safety ground, there would be noise injection, but why would anyone do that?
jd
Last edited:
Hey Guys,
Spice assumes the inductance to be linear and I have found the closer you make the device to the ideal, the closer the behavior matches what spice predicts.
The source impedance and the copper losses (DCR) really don't play much of a role in this situation, the attached picture shows the source impedance and the DCR of the chokes swept with the listed values and you can see that all 12 traces are within 2dB of each other. By the time you get the values of the Rsource or DCR to have a substantial damping effect, other issues will dominate the situation.
I also do not think core losses come into play here since at the frequency and flux levels we are dealing with the losses have to be close to zero. (not to mention many chokes use low loss materials to begin with)
This leaves me with nonlinearity as the primary reason the measured response of some devices do not match the simmed response. The two places non-linearity show up is with minimally gapped devices and at very low flux levels and it just so happens that the majority of these devices operate under those very conditions.
dave
Spice assumes the inductance to be linear and I have found the closer you make the device to the ideal, the closer the behavior matches what spice predicts.
The source impedance and the copper losses (DCR) really don't play much of a role in this situation, the attached picture shows the source impedance and the DCR of the chokes swept with the listed values and you can see that all 12 traces are within 2dB of each other. By the time you get the values of the Rsource or DCR to have a substantial damping effect, other issues will dominate the situation.
I also do not think core losses come into play here since at the frequency and flux levels we are dealing with the losses have to be close to zero. (not to mention many chokes use low loss materials to begin with)
This leaves me with nonlinearity as the primary reason the measured response of some devices do not match the simmed response. The two places non-linearity show up is with minimally gapped devices and at very low flux levels and it just so happens that the majority of these devices operate under those very conditions.
dave
Here's a quick reference or two to look at: Resonance in series-parallel circuits : RESONANCE which has several notes regarding core losses, dcr, anti-resonance, and what are some limitations on modeling in pspice.
Another good reference is a post by Paul Joppa that was posted on Audio Aslyum in which he demonstrated that even a conventional airgapped series fed IT can have wicked subsonic resonances since it's L is in series (often times) with the last C in the power supply.
Yes... you must pay some attention to resonances WHENEVER and WHEREVER you have multiple reactive components in series or parallel including power supplies and crossovers. But beating back the dragon in these real world applications has not proven to be very difficult whatsoever.
If your going to lay awake at night---worried about resonances--- you better also design your circuits with steep low pass filters--- almost all (as just one example) audio output transformers will have a resonance within the first decade of response above 20 khz. In various cases the high frequency resonance will be just one, two or three octaves above 20 khz.
Using some crude modeling techniques you can also get or see low frequency resonances say in a LC coupling circuit. And as just one small point we frequently seem to miss is.... at the subsonic frequencies where even the most simple models predict a resonance what are the real world conditions in which such a resonance would actually be excited? In some of the boogey man cases presented the resonance is nearly a full DECADE below 20 hertz.
Some of the modeling and arm waving defies common sense--- if you were going to have a plus 24 db peak at 5 hertz--- which was capable of being excited in the real world---- if such a resonance were excited it would be a very short duration resonance indeed since at the levels suggested it is most likely that the iron core itself would be saturated--- and could someone tell me what the perm of a saturated core is? What happens to the effective core losses at saturation?
Also... in real world apps... real amplifiers... if you were exciting resonances in actual playback of plus 10 to 24 dbs and higher.... I'm pretty sure you would hear it from your amps in the form of gross distortion and breakup.
The truth and reality is that many, many hundreds of DIY'ers have successfully built circuits using LC coupling---- and to date I've not heard of a single spicy hopped up resonance that drove the diy'er to lose any sleep whatsoever.
MSL
Another good reference is a post by Paul Joppa that was posted on Audio Aslyum in which he demonstrated that even a conventional airgapped series fed IT can have wicked subsonic resonances since it's L is in series (often times) with the last C in the power supply.
Yes... you must pay some attention to resonances WHENEVER and WHEREVER you have multiple reactive components in series or parallel including power supplies and crossovers. But beating back the dragon in these real world applications has not proven to be very difficult whatsoever.
If your going to lay awake at night---worried about resonances--- you better also design your circuits with steep low pass filters--- almost all (as just one example) audio output transformers will have a resonance within the first decade of response above 20 khz. In various cases the high frequency resonance will be just one, two or three octaves above 20 khz.
Using some crude modeling techniques you can also get or see low frequency resonances say in a LC coupling circuit. And as just one small point we frequently seem to miss is.... at the subsonic frequencies where even the most simple models predict a resonance what are the real world conditions in which such a resonance would actually be excited? In some of the boogey man cases presented the resonance is nearly a full DECADE below 20 hertz.
Some of the modeling and arm waving defies common sense--- if you were going to have a plus 24 db peak at 5 hertz--- which was capable of being excited in the real world---- if such a resonance were excited it would be a very short duration resonance indeed since at the levels suggested it is most likely that the iron core itself would be saturated--- and could someone tell me what the perm of a saturated core is? What happens to the effective core losses at saturation?
Also... in real world apps... real amplifiers... if you were exciting resonances in actual playback of plus 10 to 24 dbs and higher.... I'm pretty sure you would hear it from your amps in the form of gross distortion and breakup.
The truth and reality is that many, many hundreds of DIY'ers have successfully built circuits using LC coupling---- and to date I've not heard of a single spicy hopped up resonance that drove the diy'er to lose any sleep whatsoever.
MSL
Last edited:
As you and Dave mention it might be a matter of simple magnetic circuit quality. The Hammonds were modeled properly for L, R and winding capacitance but didn't show anywhere near predicted peak driven by a known and high R. I'm pretty much in line with your thinking, transformer or DC couple.
I was trying to find Paul Joppa's post on resonances in parafeed and series type circuits that he posted on Audio Aslyum quite a while ago.... too bad their search engine is about worthless.
The irony is... that a series fed IT ALSO has the necessary constituent reactive components to create a resonance...
Whereas in LC coupling you have a C in series above the L... in most series feed IT based circuits you will have a C in series below the L (on the earthy side). So you actually have an LC circuit in each case.
Conceptually--- you could think of the series fed conventional IT as a plate choke and a grid choke built on one core. The primary of the series fed IT is a "plate choke"..... in most cases not a very good one because of it's often limited L so that it does not load the tube very well... and the secondary of the series IT (and para IT as regards the secondary) is the grid choke... but, again, not a very good grid choke in most cases.
Conceptually--- you might think of using say a discrete plate choke and a grid choke as an opportunity to optimize your circuit moreso than what most conventional airgapped IT's will allow.
Also... another good resource is Voltsec's website... where he has an in depth discussion of parafeed and some additional notes on grid chokes and etc. I will post his url as soon as I can find it
MSL
The irony is... that a series fed IT ALSO has the necessary constituent reactive components to create a resonance...
Whereas in LC coupling you have a C in series above the L... in most series feed IT based circuits you will have a C in series below the L (on the earthy side). So you actually have an LC circuit in each case.
Conceptually--- you could think of the series fed conventional IT as a plate choke and a grid choke built on one core. The primary of the series fed IT is a "plate choke"..... in most cases not a very good one because of it's often limited L so that it does not load the tube very well... and the secondary of the series IT (and para IT as regards the secondary) is the grid choke... but, again, not a very good grid choke in most cases.
Conceptually--- you might think of using say a discrete plate choke and a grid choke as an opportunity to optimize your circuit moreso than what most conventional airgapped IT's will allow.
Also... another good resource is Voltsec's website... where he has an in depth discussion of parafeed and some additional notes on grid chokes and etc. I will post his url as soon as I can find it
MSL
Last edited:
How did you account for the change in inductance with applied flux? Just did a quick test and a 150hy 8ma choke with a DCR of 3400 ohms on the core size of the 156 gave me the predicted 150hy up to 8ma (20Vac excitation at 50hz) However when I measured it @ 50mv @ 100hz the inductance is 100hy. You do need to remember that this is a choke with a sizable air gap so it is approaching a linear device.
I simmed and measured a circuit with the above mentioned choke with a .1uf 1uf and 3uf coupling cap and the measured results were really close to what the sims suggested.
now consider the relationship that goes on. Perm (hence inductance) increases up until saturation then it decreases.
As frequency goes down the resonance increases the AC value which increases the perm. Also as frequency goes down, flux goes up which also increases the perm. This suggest as the resonance is starting, the inductance starts to increase. By increasing the inductance the resonant frequency goes down making everything shift down in frequency. I suspect this repeats until you hit saturation and the perm suddenly drops. By the time this happens, you are so far past your original resonant frequency that the peak never has a chance to materialize.
In other words, I suspect it is the choke nonlinearity that causes the difference between real and simmed results and actually a really nonlinear device can make the peak almost disappear completely.
One final thing of interest in this situation is that a frequency sweep on my soundcard does not show the same peaking that a signal generator and scope does... How were you measuring?
dave
I didn't. I've been horrid at keeping records of this stuff and still stumble across results on projects I completely forgot having done. My recollection is monitoring the voltage across the choke with a scope (my sound cards are inappropriate at 10k input impedance), and driving the circuit with a sound card or Beckman generator, a large resistor in series with the DUT. So it would have been low voltages, no DC bias. What you're saying though gets back to a previous discussion on AA about (some) chokes potentially modulating the low end frequency response with excitation level.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.