REW Update
- Thread starter alexc
- Start date
Help Support GroupDIY Audio Forum:
And, being prone to editorialising my own narrations 
at the relaxing evening after the day's work ...
...
I nominate REW for the award of 'most wonderful quant tech for audio to date'
It's getting a little ..'scary' accurate, in my measurement world!
Soon I have to go back to 'design' rather than just 'roughing things out' on the bench - that would be a first for me.
.. to get any more 'actual' improvements [in my my own diy audio builds]
..
I may have to make up an award 'graphic' to commemorate what REW means to me, later of 4 decades chasing audio at the 'poorman' end of town.
<pic of rew 'award' Jan 2019 - 'services to audio measurement kind'>
cue 'Dark End of the Street' as performed by Mr Gram Parsons and his band
followed by something like 'Afternoon Delight' , I think 8)
at the relaxing evening after the day's work ......
I nominate REW for the award of 'most wonderful quant tech for audio to date'
It's getting a little ..'scary' accurate, in my measurement world!
Soon I have to go back to 'design' rather than just 'roughing things out' on the bench - that would be a first for me.
.. to get any more 'actual' improvements [in my my own diy audio builds]
..
I may have to make up an award 'graphic' to commemorate what REW means to me, later of 4 decades chasing audio at the 'poorman' end of town.
<pic of rew 'award' Jan 2019 - 'services to audio measurement kind'>
cue 'Dark End of the Street' as performed by Mr Gram Parsons and his band
followed by something like 'Afternoon Delight' , I think 8)
Just a little more info on tube testing with REW ..... and 'HV Probe' stuff.
As I described, I have my 'HV Probe' with an attenuation of -26.5dB - a value chosen to offset the overall gain of the 'circuit under test' - thusly to give a reasonably 'normalised REW snapshot' - with the return around the 0dBu (nominal) - easy on the eyes.
I was testing the output of a 'gca + cf' functional block with no real 'loading effects' evident at any attenuation settings.
When I switched to a 'gca' functional block, with the higher output Z - the loading effect of attenuator set to 'minimum attenuation' becomes very noticeable. [the higher attenuation settings are, as usual, quite negligible, for loading]
SO - it is useful to have a CF at the front of the 'HV Probe' ... permanently
The super high input impedance of the CF buffers even the highest output Z of the 'circuit under test'
and the intrinsically low and 'known' THD of the CF doesn't impact typical tube measurements by much ... less than an impedance matching transformer, for example.
....
Hence, I will do a 'tube active HV Probe' unit .. going forward .. with an 11 position attenuator to really dial in the 'net gain/loss' factor as well as ensuring that loading effects stay well and truly in-the-box.
It's worthwhile to do because I already have the remote PSU completed and in use for my 'Tube Test Rig'
....
I'm thinking to try an 'Aikido' type CF circuit ... perhaps now that I've got my basic tube performance 'in the zone' I will be able to measure a difference over say a White CF circuit, which would be my usual choice.
OF course it means another 'investigation' to do ... which tube is best for a CF ?
Experience tells me that (again) the 6922 may be best ... at least for input voltage swings that don't get too large.
For measurements with several hundred volts swing ... it will need something with a really high B+ ... I think about 375V is my limit with this rig ... I would say a JJ 6V6 or even two in triode mode could be a good choice ... for a 'measurement CF'.
At least, after the CF stage, the 'attenuation' is fairly easy to make .. a combination of 'fixed' and 'variable' resistive attens in series, I would think ... and also some nfb could do the job I hope .....
.....
As soon as I finish this '6922' and 'E83F' comparison, I'm going to get right on it.
....
The other thing I need to figure out, is how to 'interpret' the changes in THD when switching from a 'higher attenuation' setting to a 'lower attenuation' setting on my passive 'HV Test Probe'.
ie. I can track the gain changes etc perfectly, but I do see a difference in THD which I can't account for.
eg. the -26.5 dB atten shows higher THD than the -37dB atten
It's not the analyser ... as far as I can tell
ie. normally if a signal is a bit too low in amplitude, then THD is reported some little amount higher, until a certain level, then no further.
It's not the 'circuit under test' it's a 'poor analyzer signal-noise ratio' kind of thing.
BUT this is something else - I think it's just that the 'lighter loading' of the attenuator when at 'high atten settings' means lower THD (despite the overall increase in level).
When using a 'lower attenuation', the loading is 'higher' and so the THD is also higher.
Once I have some more data with the 'CF buffered' HV Test Probe, I should figure it out.
Early indications are that once 'loading effects' are reduced to a minimum, the THD should be reasonably steady at differing atten settings.
Hopefully, the 'fixed high input Z' of the CF will not also increase my 'noise floor' too much!
....
Measurement is a deep, deep well .. you can drop hundreds of hours in it before hearing an echo
As I described, I have my 'HV Probe' with an attenuation of -26.5dB - a value chosen to offset the overall gain of the 'circuit under test' - thusly to give a reasonably 'normalised REW snapshot' - with the return around the 0dBu (nominal) - easy on the eyes.
I was testing the output of a 'gca + cf' functional block with no real 'loading effects' evident at any attenuation settings.
When I switched to a 'gca' functional block, with the higher output Z - the loading effect of attenuator set to 'minimum attenuation' becomes very noticeable. [the higher attenuation settings are, as usual, quite negligible, for loading]
SO - it is useful to have a CF at the front of the 'HV Probe' ... permanently
The super high input impedance of the CF buffers even the highest output Z of the 'circuit under test'
and the intrinsically low and 'known' THD of the CF doesn't impact typical tube measurements by much ... less than an impedance matching transformer, for example.
....
Hence, I will do a 'tube active HV Probe' unit .. going forward .. with an 11 position attenuator to really dial in the 'net gain/loss' factor as well as ensuring that loading effects stay well and truly in-the-box.
It's worthwhile to do because I already have the remote PSU completed and in use for my 'Tube Test Rig'
....
I'm thinking to try an 'Aikido' type CF circuit ... perhaps now that I've got my basic tube performance 'in the zone' I will be able to measure a difference over say a White CF circuit, which would be my usual choice.
OF course it means another 'investigation' to do ... which tube is best for a CF ?
Experience tells me that (again) the 6922 may be best ... at least for input voltage swings that don't get too large.
For measurements with several hundred volts swing ... it will need something with a really high B+ ... I think about 375V is my limit with this rig ... I would say a JJ 6V6 or even two
in triode mode could be a good choice ... for a 'measurement CF'.At least, after the CF stage, the 'attenuation' is fairly easy to make .. a combination of 'fixed' and 'variable' resistive attens in series, I would think ... and also some nfb could do the job I hope .....
.....
As soon as I finish this '6922' and 'E83F' comparison, I'm going to get right on it.
....
The other thing I need to figure out, is how to 'interpret' the changes in THD when switching from a 'higher attenuation' setting to a 'lower attenuation' setting on my passive 'HV Test Probe'.
ie. I can track the gain changes etc perfectly, but I do see a difference in THD which I can't account for.
eg. the -26.5 dB atten shows higher THD than the -37dB atten
It's not the analyser ... as far as I can tell
ie. normally if a signal is a bit too low in amplitude, then THD is reported some little amount higher, until a certain level, then no further.
It's not the 'circuit under test' it's a 'poor analyzer signal-noise ratio' kind of thing.
BUT this is something else - I think it's just that the 'lighter loading' of the attenuator when at 'high atten settings' means lower THD (despite the overall increase in level).
When using a 'lower attenuation', the loading is 'higher' and so the THD is also higher.
Once I have some more data with the 'CF buffered' HV Test Probe, I should figure it out.
Early indications are that once 'loading effects' are reduced to a minimum, the THD should be reasonably steady at differing atten settings.
Hopefully, the 'fixed high input Z' of the CF will not also increase my 'noise floor' too much!
....
Measurement is a deep, deep well .. you can drop hundreds of hours in it before hearing an echo
ruffrecords
Well-known member
Regarding the difference you see in THD with different probe settings, just about all tube circuits produce more distortion with a lower value load. Exactly how this varies with load depends on the particular design but it is reasonable to assume that the probe lowers the value of the load so the distortion increases so what you measure is a worst case.
Cheers
Ian
Cheers
Ian
Thanks Ian - that does agree with what I am measuring.
....
I'm looking forward to exploring the CF circuit as an input stage to my 'tube active measurement box' MK2
I'm very interested in what perf one can get from a 'power pentode' wired for triode mode .. as a 'single stage' CF and also, just for the learning ... what a pair can do in a push-pull CF config .... firstly a White CF and if I get any joy .. an Aikido or Broskie type.
It will have to wait for my Mk2 remote PSU to get the real voltages going ... +475V and such [ as high a B+ as I am game to use]
I *do* like to measure plates in power stages
....
So far, the 6922 as single CF is working out really well ... for signals to the grid of up to 60Vpp or so (B+ 235V)
I have almost done the E83F-triode-mode gca stage measurements ... they are similar to the 6922 gca stage measurements
ie., for same input voltage, bias and plate load, each has very similar gain and signal-hum-margin with the same CF as load.
.... just need to nail down a little excess 100Hz I'm newly seeing .. need to rewire a socket, I think .
...
And also, just 'driving down' B+ for each, to find a reasonable compromise with the increase in THD (as B+ decreases).
6922 is at +235V supply and the E83F-triode-mode is at +315V supply and they look pretty similar
....
I'm looking forward to exploring the CF circuit as an input stage to my 'tube active measurement box' MK2
I'm very interested in what perf one can get from a 'power pentode' wired for triode mode .. as a 'single stage' CF and also, just for the learning ... what a pair can do in a push-pull CF config .... firstly a White CF and if I get any joy .. an Aikido or Broskie type.
It will have to wait for my Mk2 remote PSU to get the real voltages going ... +475V and such [ as high a B+ as I am game to use]
I *do* like to measure plates in power stages
....
So far, the 6922 as single CF is working out really well ... for signals to the grid of up to 60Vpp or so (B+ 235V)
I have almost done the E83F-triode-mode gca stage measurements ... they are similar to the 6922 gca stage measurements
ie., for same input voltage, bias and plate load, each has very similar gain and signal-hum-margin with the same CF as load.
.... just need to nail down a little excess 100Hz I'm newly seeing .. need to rewire a socket, I think .
...
And also, just 'driving down' B+ for each, to find a reasonable compromise with the increase in THD (as B+ decreases).
6922 is at +235V supply and the E83F-triode-mode is at +315V supply and they look pretty similar
Then you need a CF that is capable of swinging 1000V+, not a minor endeavour.alexc said:I *do* like to measure plates in power stages
abbey road d enfer said:
Of course, I will retain my 'passive hv probe' scheme .. particulary for the the really big stuff .. 450V plate swings and such.
There, the 'attenuation factor' is much larger, meaning my 'loading effects' are reasonably minimal ... whilst still not attenuating below my 'measurement channel' hum floor.
So
-> an 'active CF HV probe' for as far as I can 'swallow input swing' for a B+ 475V - maybe 125Vpp or so ?
-> a 'passive HV probe' for the signal beyond that 125Vpp .. 450Vpp
I think it's a nice challenge
There, the 'attenuation factor' is much larger, meaning my 'loading effects' are reasonably minimal ... whilst still not attenuating below my 'measurement channel' hum floor.
So
-> an 'active CF HV probe' for as far as I can 'swallow input swing' for a B+ 475V - maybe 125Vpp or so ?
-> a 'passive HV probe' for the signal beyond that 125Vpp .. 450Vpp
I think it's a nice challenge
And I don't have much idea of what some power mosfets can do in this kind of application,
... but I have some bunch of affordable IRF B20N50K devices, which I'm using in HV regs ... I'm having some explore time with .. in very simple circuits ... I like the 'fet to tube' conversion kind of thing in these basic functional blocks.
... but I have some bunch of affordable IRF B20N50K devices, which I'm using in HV regs ... I'm having some explore time with .. in very simple circuits ... I like the 'fet to tube' conversion kind of thing in these basic functional blocks.
.. and here's the 'normalized' E83F scenario for 1% THD .... using the 6922 CF stage before the 'HV Probe' attenuator
At +315V plate supply, 47K plate load resistor ... with 2Vpp input giving 47.3Vpp output of gca (+27.5dB)
This can be a direct comparison now to the '6922 gca' which is my benchmark for this kind of config
...
Comparing the E83F and 6922 in these 'normalized' views,
.. the 6922 is a smidgeon better .... a fraction of a dB more gain ... a tiny increment less THD with same 'hum floor' and 'freq resp' under these same loading conditions.
The E83F-triode-mode - is more or less identical to the 6922 except for it's much lower supply voltage ... +245V vs +315V for the pentode.
Each has a typical 'signal margin over hum' of +116dB at 1% THD (1KHz sine) .... rising to +126dB at 3% THD
....
Each of these tubes can do a very nice job Not bad really ... for the E83F ... to be close-run second to the 6922 .... so far ;D
Beyond this very simple comparison, each goes on to diverse applications ....
- the 6922 with it's 2x triodes in one envelope for low cost .. with 365mA heater at 6.3V .. finds wide usage in 'front end' s
- the E83F for more 'output' oriented duties for 'more cost' ...with 300mA heater at 6.3V ... 'single ended driver' s
And so it goes! Until I wire up my 'av7 style' socket .. and can bring that lot into the fray, that is.
At +315V plate supply, 47K plate load resistor ... with 2Vpp input giving 47.3Vpp output of gca (+27.5dB)
This can be a direct comparison now to the '6922 gca' which is my benchmark for this kind of config
...
Comparing the E83F and 6922 in these 'normalized' views,
.. the 6922 is a smidgeon better .... a fraction of a dB more gain ... a tiny increment less THD with same 'hum floor' and 'freq resp' under these same loading conditions.
The E83F-triode-mode - is more or less identical to the 6922 except for it's much lower supply voltage ... +245V vs +315V for the pentode.
Each has a typical 'signal margin over hum' of +116dB at 1% THD (1KHz sine) .... rising to +126dB at 3% THD
....
Each of these tubes can do a very nice job
Not bad really ... for the E83F ... to be close-run second to the 6922 .... so far ;DBeyond this very simple comparison, each goes on to diverse applications ....
- the 6922 with it's 2x triodes in one envelope for low cost .. with 365mA heater at 6.3V .. finds wide usage in 'front end' s
- the E83F for more 'output' oriented duties for 'more cost' ...with 300mA heater at 6.3V ... 'single ended driver' s
And so it goes! Until I wire up my 'av7 style' socket .. and can bring that lot into the fray, that is.
Attachments
Next task is to get a 'power pentode' going as a cathode follower ... to get a feel for the THD one can expect at various input voltage swings - as some pre-work for my 'cathode follower input buffer' option for my Mk2 'HV Test Probe' .
....
And while I'm getting that going, I'm going to have a look at the 'spread' of performance with some more 6922, then with some of the 'extended family' ... including some nos 'greats' and some russian equivalents ...
And of course, I'll do some more 'historical' runs and update my data on my *fav standout performer*s in this area ...
- the Orange86 and the New York Dave One Bottle pres.
Of course, I can only speculate what my other 'great', the Gyraf G9 would return in this 'test environment'
My G9 is pretty good as built .. but I can't help but wonder just how good it could get with an optimal psu
....
And while I'm getting that going, I'm going to have a look at the 'spread' of performance with some more 6922, then with some of the 'extended family' ... including some nos 'greats' and some russian equivalents ...
And of course, I'll do some more 'historical' runs and update my data on my *fav standout performer*s in this area ...
- the Orange86 and the New York Dave One Bottle pres.
Of course, I can only speculate what my other 'great', the Gyraf G9 would return in this 'test environment'
My G9 is pretty good as built .. but I can't help but wonder just how good it could get with an optimal psu
ruffrecords
Well-known member
There are two main current production sources of the 6922. One is the 6922EH by Electroharmonix and the other is the JJE88CC Back when I first started using the 6922 I tested a few of each. Generally they were both OK but the JJ types had some unusual harmonic harmonic structure. The EH ones had mostly H2 with the other harmonics steadily decreasing. All the JJ ones had quite a loss less H2, more H3 than the EH ones and thereafter steadily declining harmonics. I also had a couple of noisy JJ ones so in the end I 'standardised' on the EH ones.
Cheers
Ian
Cheers
Ian
That's interesting Ian ... I only have modern in 'eh6922' ... I've always felt they were better than I expected.
I do have more of the '6922' family than I need, but if I was going to try another modern one, it would be the JJ
But I don't think I could bare a 'more h3' tube anymore ... I know, h3 is just as important in it's way .. what with the 'bright' and 'cut thru' qualities.
It's just that I'm a relatively simple person .. h2 dominating is not so easy to come across 'in the wild' and I just revel in it.
I should say, that it can get too much at times ... like a 'resonator' gone a bit too far (long cardboard tube + kids)
But ... yep .. I am interested to see the 'spread' of results for the '6922' family that I have 'in stock'.
...
I find it amazing that once I'd given my best 'knowledge' plays *and* all the 'REW magic' ... together .. giving me my 'best yet' results ... that they wound up neatly 'normalized' for the notional '1% thd +n (headline)' figure *and* centred about '0dBu'
... before I even knew what is 'really' ... measurement detail wise ..
But yes .. it all pops out nicely ready for casual view-ment.
As the 'Small Faces' said (and oft afterwards) "Its All Too Beautiful"
I do have more of the '6922' family than I need, but if I was going to try another modern one, it would be the JJ
But I don't think I could bare a 'more h3' tube anymore ... I know, h3 is just as important in it's way .. what with the 'bright' and 'cut thru' qualities.
It's just that I'm a relatively simple person .. h2 dominating is not so easy to come across 'in the wild' and I just revel in it.
I should say, that it can get too much at times ... like a 'resonator' gone a bit too far (long cardboard tube + kids)
But ... yep .. I am interested to see the 'spread' of results for the '6922' family that I have 'in stock'.
...
I find it amazing that once I'd given my best 'knowledge' plays *and* all the 'REW magic' ... together .. giving me my 'best yet' results ... that they wound up neatly 'normalized' for the notional '1% thd +n (headline)' figure *and* centred about '0dBu'
... before I even knew what is 'really' ... measurement detail wise ..
But yes .. it all pops out nicely ready for casual view-ment.
As the 'Small Faces' said (and oft afterwards) "Its All Too Beautiful"
And with that, I'll be signing off for a while ... some 'fire warnings' .. abroad in the la nd ...
Hopefully will be 'nothing' over the next couple of days .. but elsewise I'm glad to 'have a plan'.
....
The ire-net was mostly gone for a day or so .. I think it will shutoff again soon . ... I have 4g connects to the local base-station .. but the smoke is pretty thick now.
One can prolly expect the power to go off quite soon too. Water usually follows after that, if things get fully dire.
When it goes 'orange-brown all around' ... it's time to head for the coastline ... fortunately I can cycle there pretty easy.
Way down here in Southern Tas, it's reasonably sane weather - compared to some mainland Aust regions - and it's still quite early in the 'pre- fire season'.
As I once said before, it 'doesn't take much' ..... etc.
bfn
Hopefully will be 'nothing' over the next couple of days .. but elsewise I'm glad to 'have a plan'.
....
The ire-net was mostly gone for a day or so .. I think it will shutoff again soon . ... I have 4g connects to the local base-station .. but the smoke is pretty thick now.
One can prolly expect the power to go off quite soon too. Water usually follows after that, if things get fully dire.
When it goes 'orange-brown all around' ... it's time to head for the coastline ... fortunately I can cycle there pretty easy.
Way down here in Southern Tas, it's reasonably sane weather - compared to some mainland Aust regions - and it's still quite early in the 'pre- fire season'.
As I once said before, it 'doesn't take much' ..... etc.
bfn
Luck for me, the fires turned away with a wind change .... they reached about 10km or so away from me.
Today is super nice again ... regular 25C temperatures, the UV 'burn' is way down, gentle sea breezes and all the smoke from yesterday .. gone!
It went from 'eve of destruction' to 'pleasant valley sunday' overnight.
Still .. the fires are all going and this weekend is predicted to be a shocker ... hopefully my area will just bypass the drama .. as it has since the mid-60s .. which was the last time my township was severe with fire.
So far, so good!
....
I put together today a 6V6-triode-mode cathode follower ... to test what range of input signal I can get with low thd ... +325V supply, and the cathode is expected to be around +180V or so, for 27mA of current for around 8.8W of plate dissipation.
It's self biased at around +27V and (1K+5K3) resistors (high wattage!) in the cathode circuit, bypassed with 100uF electro.
Hopefully, I can get something like 200Vpp input capability at low THD ... I'll then add some negative feedback .. to give some attenuation as well as reduction in distortion.
Hopefully I can get some numbers tomorrow.
Today is super nice again ... regular 25C temperatures, the UV 'burn' is way down, gentle sea breezes and all the smoke from yesterday .. gone!
It went from 'eve of destruction' to 'pleasant valley sunday' overnight.
Still .. the fires are all going and this weekend is predicted to be a shocker ... hopefully my area will just bypass the drama .. as it has since the mid-60s .. which was the last time my township was severe with fire.
So far, so good!
....
I put together today a 6V6-triode-mode cathode follower ... to test what range of input signal I can get with low thd ... +325V supply, and the cathode is expected to be around +180V or so, for 27mA of current for around 8.8W of plate dissipation.
It's self biased at around +27V and (1K+5K3) resistors (high wattage!) in the cathode circuit, bypassed with 100uF electro.
Hopefully, I can get something like 200Vpp input capability at low THD ... I'll then add some negative feedback .. to give some attenuation as well as reduction in distortion.
Hopefully I can get some numbers tomorrow.
ruffrecords
Well-known member
The thing about a cathode follower is the cathode follows the grid so the signal level can be very large without suffering grid current. So you don't need to bias to a high current as you would with a grounded cathode stage.
Cheers
Ian
Cheers
Ian
So far, I have the 6v6-triode cathode follower going, but not too well.
I always get my operating point wrong for CFs 8) .. forgetting to subtract the cathode V from the plate V on the sheets.
I measure B+325V, grid-to-cathode voltage of +17V (bias) with cathode at +112V ... my 1K+6K4 cathode string has +95V across them.
Which does indeed 'measure up' on the data sheet ...
....
But, I wanted the cathode to be at around +160V or half the supply .. and at quite a few more mA ... 25mA instead of the measured 17mA or so .. which is wrong by about '100 supply volts'
Still have some adjustments to make
I can increase B+ to 350V, and decrease the 'self bias resistor' part of the cathode string .. to coax more current .. and increase a bit the cathode load resistor ... for higher V ... and as well as give it a 'defined load' .. and I may need a 'stiffer grid circuit resistance' .. using 560K right now .. prolly may need to go up to 1M .. perhaps, that is.
That may all give me some +135V or so at the cathode .. I think that should give 120Vpp or so capability.
..
I am getting quite a lot of loss of gain right now, which I should be able to fix .. but it didn't seem to be a 'loading' issue. Instead of around gain of x0.93 it's more like x0.77, which is excessive.
Also, the hum floor is a bit higher than in the 6922 cf case ... -111dBu vs -116dBu .. it's a 100Hz spike of 6dB ... can't seem to shake. Usually that is 'supply decoupling' issue ...
ie. add a 'dropper resistor' into a stage's B+ line, with an electro cap to ground (47uF 450V)
and it usually goes away ... but not this time. We shall see, as I go forward.
...
Right now, THD is quite good ... I am feeding the 6v6-triode cf from the e83f gca stage of previous ... basically 46Vpp at 1%.
That returns as around 0.76% THD at the output of the 6v6-triode cf .. when I increase the input signal to 'compensate for the lossy-ness', that goes to 1.0% and with +110dB of signal-margin-over-hum.
Which is OK but not flash ... in line with my very lousy operating conditions!
....
The other thing is that a CF made with a power pentode (in triode mode) should have the advantage (over say a 6922 cf) for 'higher input swing' capability, hopefully at low noise.
It does 'sort of' , but to test properly, I need now to rig up some more 'clean gain' for it .. in order to verify it's low thd and large swing capability, as a low Z output.
Something like 100Vpp to 200Vpp at as clean as I can make. Usually I use a step up transformer, but I should have some tube section standing by .. in this 'Tube Test Rig'.
If I can get that missing '6dB' of signal-margin-over hum at low-THD .. with a stronger input signal, then all is well ... as a starting point.
....
AND, in parallel to the above, I'm doing the same thing again, right next to the octal .. but in a noval 600mA 6.3 heater 'single triode 8.5W max plate' 6s4a
.. which is a fav of mine
Similar to the 6v6-triode cf now and probably better. It's early days .. just 1/2 hr on the clock for the 6V6gt !
The other thing I'll do, is try a 6L6gt in that octal position too ... just for some additional data points
I always get my operating point wrong for CFs 8) .. forgetting to subtract the cathode V from the plate V on the sheets.
I measure B+325V, grid-to-cathode voltage of +17V (bias) with cathode at +112V ... my 1K+6K4 cathode string has +95V across them.
Which does indeed 'measure up' on the data sheet ...
....
But, I wanted the cathode to be at around +160V or half the supply .. and at quite a few more mA ... 25mA instead of the measured 17mA or so .. which is wrong by about '100 supply volts'
Still have some adjustments to make
I can increase B+ to 350V, and decrease the 'self bias resistor' part of the cathode string .. to coax more current .. and increase a bit the cathode load resistor ... for higher V ... and as well as give it a 'defined load' .. and I may need a 'stiffer grid circuit resistance' .. using 560K right now .. prolly may need to go up to 1M .. perhaps, that is.
That may all give me some +135V or so at the cathode .. I think that should give 120Vpp or so capability.
..
I am getting quite a lot of loss of gain right now, which I should be able to fix .. but it didn't seem to be a 'loading' issue. Instead of around gain of x0.93 it's more like x0.77, which is excessive.
Also, the hum floor is a bit higher than in the 6922 cf case ... -111dBu vs -116dBu .. it's a 100Hz spike of 6dB ... can't seem to shake. Usually that is 'supply decoupling' issue ...
ie. add a 'dropper resistor' into a stage's B+ line, with an electro cap to ground (47uF 450V)
and it usually goes away ... but not this time. We shall see, as I go forward.
...
Right now, THD is quite good ... I am feeding the 6v6-triode cf from the e83f gca stage of previous ... basically 46Vpp at 1%.
That returns as around 0.76% THD at the output of the 6v6-triode cf .. when I increase the input signal to 'compensate for the lossy-ness', that goes to 1.0% and with +110dB of signal-margin-over-hum.
Which is OK but not flash ... in line with my very lousy operating conditions!
....
The other thing is that a CF made with a power pentode (in triode mode) should have the advantage (over say a 6922 cf) for 'higher input swing' capability, hopefully at low noise.
It does 'sort of' , but to test properly, I need now to rig up some more 'clean gain' for it .. in order to verify it's low thd and large swing capability, as a low Z output.
Something like 100Vpp to 200Vpp at as clean as I can make. Usually I use a step up transformer, but I should have some tube section standing by .. in this 'Tube Test Rig'.
If I can get that missing '6dB' of signal-margin-over hum at low-THD .. with a stronger input signal, then all is well ... as a starting point.
....
AND, in parallel to the above, I'm doing the same thing again, right next to the octal .. but in a noval 600mA 6.3 heater 'single triode 8.5W max plate' 6s4a
.. which is a fav of mine
Similar to the 6v6-triode cf now and probably better. It's early days .. just 1/2 hr on the clock for the 6V6gt !
The other thing I'll do, is try a 6L6gt in that octal position too ... just for some additional data points
ruffrecords
Well-known member
The thing about a cathode follower is the cathode follows the grid so the signal level can be very large without suffering grid current. So you don't need to bias to a high current as you would with a grounded cathode stage.
Cheers
Ian
Cheers
Ian
I do follow you Ian ... at the 'first level' of cf understanding, as it were - so a moderate current is ok .. I don't really need to overdo it ( current wise) to still have a decent input signal handling .. at a reasonable thd into a reasonable load ?
Based on previous stuff, I think it's fair to say I've had better result (so far) with say a 6SN7 and/or 6SL7 in this regard ...
But it's all experience - so I'll do some more thinking
Based on previous stuff, I think it's fair to say I've had better result (so far) with say a 6SN7 and/or 6SL7 in this regard ...
But it's all experience - so I'll do some more thinking
