Along with a schematic of a HarmonKardon250 I have changed the input stage just ahead of the split load phase inverter. In the original schematic the first half of the 6CG7 derived it's plate voltage by conecting it's plate to the cathode of the phase inverter stage through a resistor. I also did away with the coupling cap between the first stage and the PI by choosing the new Q voltage on the first 1/2 of the 6CG7 to be B+/3 as suggested in many of the threads I have read. I think these changes are good for the overall operation of the amp, but I would very much appreciate some feedback especially concerning NFB. In the original schematic the screen grids of the output tubes were tied to thier own 310v source, but I would like to operate in Ultralinear(as shown in my poorly drawn schematic). By going UL, i understand that to be a form of NFB in itself,also the first stage is under current injected feedback(correct)before I even bring any output voltage back to the stage. Do you think there is enough NFB already or do I need to bring some back from the secondary(excuse the poorly drawn Transformer)? Also please point out any obvious mistakes or perhaps it should be canned and start all over. Thanks for newbiehelp
Thanks SY. When I plotted the load line for the 1st half of the 6CG7 tube i did see that it was almost completely horz. but I was using the values that the tubes data sheet said would work and figured "hey they know better than me". As far as gain for the amp, if I have this correct, becuase my output tubes need 33 volts(RMS) swing each for full ouput, the output from the 1st stage should be slightly greater than that value because the PI will lower the gain less than unity? Crowhurst mentions that the grid resistors for the stage following a Splitload PI shouldn't be more than 250K, yet the HK amp had values of 330K. Does disaster loom at a value of 330K? I see the obvious problem of the cap across the 22k in the PI. I was confused between two different designs so it will be dropped.
So 360ohms in the cathode with an anode value of 36K will get me to Q of 5.6ma at 100volts with the grid biased at -2 volts. I see(or at least think I see) that I am going to have trouble driving the output tubes to full output once NFB is introduced because at a 2 volt swing in grid voltage I only have a peak plate voltage change of approx 45 volts each way. I see now why Byrinth recomended using the "concertina"only on small 10-15 watt amps. I think Byrinth would call this design "Bad engineering"
Thanks, I needed to discover this.Back to the Crowhurst articles with a new understanding.
So 360ohms in the cathode with an anode value of 36K will get me to Q of 5.6ma at 100volts with the grid biased at -2 volts. I see(or at least think I see) that I am going to have trouble driving the output tubes to full output once NFB is introduced because at a 2 volt swing in grid voltage I only have a peak plate voltage change of approx 45 volts each way. I see now why Byrinth recomended using the "concertina"only on small 10-15 watt amps. I think Byrinth would call this design "Bad engineering"
Thanks, I needed to discover this.Back to the Crowhurst articles with a new understanding.
Horizontal is good, but low voltages and currents on 6CG7 (and the sister version, 6SN7) aren't.
The resistors following the split-load (i.e., the grid resistors of the output tubes) are really sized more to accomodate the output tube requirements. From the driver's point of view, the bigger the better, but the output stage wants them small because of grid leakage currents. In cathode bias configuration, a 6L6GC allows up to 500K for that resistor, so you're completely safe.
There are some ways you can get more gain in that first stage: you can cascode (a FET on the bottom works well), you can use a higher gain tube (I'd look at 12AT7/ECC81), you can even use (gasp!) a pentode. But the limitation is sure to be the rather limited swing of the split-load.
For tubes like the 6L6 which need a lot of drive, I think cascaded differential amps are all in all the best solution. Crowhurst didn't have good, practical constant current sources/sinks at hand, but in 2006, we do, so that tilts the field of engineering choices.
The resistors following the split-load (i.e., the grid resistors of the output tubes) are really sized more to accomodate the output tube requirements. From the driver's point of view, the bigger the better, but the output stage wants them small because of grid leakage currents. In cathode bias configuration, a 6L6GC allows up to 500K for that resistor, so you're completely safe.
There are some ways you can get more gain in that first stage: you can cascode (a FET on the bottom works well), you can use a higher gain tube (I'd look at 12AT7/ECC81), you can even use (gasp!) a pentode. But the limitation is sure to be the rather limited swing of the split-load.
For tubes like the 6L6 which need a lot of drive, I think cascaded differential amps are all in all the best solution. Crowhurst didn't have good, practical constant current sources/sinks at hand, but in 2006, we do, so that tilts the field of engineering choices.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.