Wow, thats just what i'm after. I have no B blaster, but perhaps I could come up with something. Gosh, those laminations are the cleanest I have seen! I'm real close to pulling the trigger on those, Along with the Edcor PT you suggested. The savings might pay for the remainer of the amp parts. I know the Edcors are pretty good, but I just don't want to give up any SQ on this purchase. If I understand, I think Eli , is suggesting that more expensive OPT would make no audible difference with this topology. Is that correct?
TubeMack:
You could sand them with 320 emery paper, but it's hard to get in the nooks & crannies by the mounting feet and screw holes. You just need a real light sanding to dull the finish before applying the black.
Any chance you could take a crack at modding the schema for the 6GK5 based on Eli's input above? Throw it up here, and it will get dialed in correctly with enough input from the experts.
You could sand them with 320 emery paper, but it's hard to get in the nooks & crannies by the mounting feet and screw holes. You just need a real light sanding to dull the finish before applying the black.
Any chance you could take a crack at modding the schema for the 6GK5 based on Eli's input above? Throw it up here, and it will get dialed in correctly with enough input from the experts.
No.
For perspective it is sometimes handy to look at things in their limits. If you have a very high load impedance, the voltage swing peak-peak will be high, but the current swing low before the tubes saturate/cut-off. If you have a very low impedance there will be enough current swing between Vg1 = 0 and cut-off, but very little output voltage swing. Thus the optimum lies somewhere inbetween, but not an arithmetic average, because distortion rises quite steeply at low load impedance.
Going to practical figures in line with this thread, fig. 5-3 in that GEC book on Amplifier Design mentioned previously is a good example. Not wishing to engage in copyright matters, I will list figures from that graph, for output and distortion against load. If you care you can draw graphs through these points on linear graph paper (smooth graphs; the scale was not large enough to read accurately):
RL(ohm): 3K---4K---4K5---5K----6K----7K----8K-----9K----10K
Pout(W): 48---53---54----53----49----46----43-----38-----35
D(%): .....4---2,7--2,2---1,6----0,9---0,5--0,45----0,45---0,45
These are for p.p. KT88s with 40% taps, h.t. = 425V, auto bias (thus actual cathode - anode and G2 voltages = 345V).
You will now notice that 'optimal load' is a judgement call; maximum output occurred at higher distortion. In the above design 6K was chosen; I would have used 7K as the drop in output is negligible compared to the drop in distortion. A 6,6K is a popular choice. You can also read this data in conjunction with the aforementioned fig. 1-1 in that other reference, regarding %G2 tap vs. other parameters.
I repeat that this represents a general pattern for beam tubes of the KT** family (including 6L6GC). For pentodes like EL34/EL84 the output power vs. distortion behaviour is slightly different though of similar nature.
Hope this gives some clarity of the broad picture regarding UL.
Thanks Johan. This gives lots of clarity.
Those values imply that a 5K/6.6K may be a better choice, if one wants to sacrifice a few watts for lower distortion. Very interesting. The 5K would sacrifice 1W for half the distortion of the 4K...
Last edited:
Not quite, the better the "iron", the better the amp will be. However, you will spend a good deal more than Edcor costs to realize a small improvement. Remember, GNFB has a homogenizing effect. Edcor stuff has been shown to work in Williamson style circuitry. Mullard style places smaller demands on the O/P trafos than Williamson style does. Why spend more than is necessary to obtain a highly satisfactory result?
On the SQ front, do you have a good o'scope to permit fine tuning phase compensation? If not, a brute force method for phase compensation that works surprisingly well is available.
"On the SQ front, do you have a good o'scope to permit fine tuning phase compensation? If not, a brute force method for phase compensation that works surprisingly well is available"
Eli, I have a brand new BK 30MHZ analog scope, and a new Rigol 100MHZ DSO with FFT. I have also just purchased a HP 339A as of 3 hours ago. I already had a Hakko 936, and a Fluke DMV, so I'm looking pretty good for this project I guess.
Yesterday I picked up a old large dual gang Vari Cap for Comp. tunning. I hope it will work. What part would be best to tune resistance?
Eli, I have a brand new BK 30MHZ analog scope, and a new Rigol 100MHZ DSO with FFT. I have also just purchased a HP 339A as of 3 hours ago. I already had a Hakko 936, and a Fluke DMV, so I'm looking pretty good for this project I guess.
Yesterday I picked up a old large dual gang Vari Cap for Comp. tunning. I hope it will work. What part would be best to tune resistance?
Last edited:
Hi Johan P.,
was there any bias current value mentioned for the KT88 on the OPT input impedance evaluation in that GEC book on Amplifier Design?
I guess it's not a very big deal if we get 50 or 60 Watts as it's only 0.8 dB of sound pressure change and virtually inaudible for most human ears else one isn't ABX'ing forth and back...
Cheers Michael
was there any bias current value mentioned for the KT88 on the OPT input impedance evaluation in that GEC book on Amplifier Design?
I guess it's not a very big deal if we get 50 or 60 Watts as it's only 0.8 dB of sound pressure change and virtually inaudible for most human ears else one isn't ABX'ing forth and back...
Cheers Michael
Last edited:
Remember, the phase compensation cap. is in parallel with the NFB resistor that connects the O/P trafo to the point where the voltage amplifier bias setup and the small 2nd NFB resistor, which goes to ground, meet. You need to think the NFB resistor pair out carefully. Too high a value in the grounded part will alter the bias setup. I would not go above 22 Ω. After deciding the grounded part's value, use the % of the O/P voltage to be fed back to compute the larger resistor's value. This yet another voltage divider situation.
Speaking of NFB, make certain that the signal at the speaker is in phase with the signal at the amp's I/P. If phase inversion occurs, feedback will be positive and all Hell can break loose.
reconsidering the Hashimoto HW-60-5.....
Eli: I just compared the specs of the Hashimoto HW-100 and the HW-60 and they don't appear to be too far apart.
HW-100-5 (100W at 30 hz)
Primary inductance 200H
unbalanced 125H, DC=5.5 ma
Primary max DC current 300 ma (2 valves)
Primary R 117 ohms
HW-60-5 (60W at 30 hz)
Primary inductance 180H
unbalanced 120H, DC=4 ma
Primary max DC current 250 ma (2 valves)
Primary R 124 ohms
Does the HW-60 still appear to "lack heft" for this topology based on the specs above? They are by far the largest iron in my parts bin. My original intent was a triode KT88 design with these, although this thread has caught my interest.
Eli: I just compared the specs of the Hashimoto HW-100 and the HW-60 and they don't appear to be too far apart.
HW-100-5 (100W at 30 hz)
Primary inductance 200H
unbalanced 125H, DC=5.5 ma
Primary max DC current 300 ma (2 valves)
Primary R 117 ohms
HW-60-5 (60W at 30 hz)
Primary inductance 180H
unbalanced 120H, DC=4 ma
Primary max DC current 250 ma (2 valves)
Primary R 124 ohms
Does the HW-60 still appear to "lack heft" for this topology based on the specs above? They are by far the largest iron in my parts bin. My original intent was a triode KT88 design with these, although this thread has caught my interest.
... for a small increment of time ... after which you will have eerie silence. All will be dead-quiet, because the amplifier would have blown, or quite more likely, your loudspeakers .... 😱 😡
(sorry, Eli - couldn't resist)
Repeat, repeat, 60 W. "iron" lacks the magnetic headroom for use with UL KT88s, in a Mullard circuit. Also, the end to end impedance is too high. You can use "exolinear" topology and full pentode mode KT88s. Then, the O/P trafo is not subject to a LF error correction signal and you can regulate g2 B+, while cranking anode B+ up. Send "Bandersnatch" a PM about this and a hybrid cascode using 6GK5s, in the LTP.
Quite - sorry I did not give full data previously.
But now my face is RED. GEC has betrayed me!! Reading in the text again, their values do not quite gel, so I will give the lot. They appear to have allowed for no/full signal supply voltages without telling (how considerate of them). Transformer is 500-0-500V, rectifier 5U4. (This will not give a no-signal d.c. of 500V; quite higher. Go figure own transformer, GZ34, whatever.)
So please; CORRECTION to my previous voltage figures. The no-load h.t. is 500V; 425V is the actual cathode-plate/screen voltage. Separate cathode resistors (bypassed) of 525 ohm each. Ia + Ig2 = 2 x 87mA no signal; or = 2 x 100mA full signal. Vg = -50V approximately - this varies with signal. Read that as the cathode bias drop over each 525 ohm. RL(a-a) = 5K; Pout = 50W; series G2 resistors = 100 ohm each to 43% UL taps. D% = 0,2% at full output (that is for the complete amplifier shown there, moderate global NFB. Circuit not particularly special; ECC82 as drivers, ECC83 as paraphase first stage. Not my preferred choice, but never mind).
Hope this is NOW correct.
No favor found here too. AAMOF it sucks! With good reason, paraphase splitters are not in vogue. Using the non-linear 12AU7, no thank you. Restrict that type to cathode follower and "concertina" phase splitter duty.
-----Repeat, repeat, 60 W. "iron" lacks the magnetic headroom for use with UL KT88s, in a Mullard circuit. Also, the end to end impedance is too high. You can use "exolinear" topology and full pentode mode KT88s. Then, the O/P trafo is not subject to a LF error correction signal and you can regulate g2 B+, while cranking anode B+ up. Send "Bandersnatch" a PM about this and a hybrid cascode using 6GK5s, in the LTP.
Eli, those OPT's are 5k a-a are they not? Also, if the load is too high( numerically ) for U-L use it is even worse for pentode mode. U-L lowers the knee, remember?
cheers,
Douglas
Eli, those OPT's are 5k a-a are they not? Also, if the load is too high( numerically ) for U-L use it is even worse for pentode mode. U-L lowers the knee, remember?
cheers,
Douglas
Doug,
It's a matter of step down ratio and voltage swing. A "tall" B+ rail and regulated g2 B+ work around that difficulty. Look at TDSL's KT88 application data, along with the KT88 data sheet. Consider a 500 V. anode reading, a regulated 300 V. g2 supply, and IB = 70 mA. as the operating point. Bias would be "fixed". That doesn't look too shabby to me.
oh my.. those Hashimoto costs $$$€€€... 🙁
Another alternative could perhaps be Lundahl LL1679/PP (C-cores)...
105 W @ 30 Hz
Primary inductance 150 H (at what current??)
Primary impedance 4,5 kOhm
UL @ 37%
Primary RDC 160 Ohm a-a
Weight 2.5 Kg each
Couldn't find max Iprim.
asked them before Christmas how much they are and would be ~1570 SEK (Swedish Krona) each excl. VAT
@ Johan, thanks btw for the clarification!
Cheers Michael
Another alternative could perhaps be Lundahl LL1679/PP (C-cores)...
105 W @ 30 Hz
Primary inductance 150 H (at what current??)
Primary impedance 4,5 kOhm
UL @ 37%
Primary RDC 160 Ohm a-a
Weight 2.5 Kg each
Couldn't find max Iprim.
asked them before Christmas how much they are and would be ~1570 SEK (Swedish Krona) each excl. VAT
@ Johan, thanks btw for the clarification!
Cheers Michael
Last edited:
Okay, I've drawn up what I understood from this very non-linear conversation. Let me know what I screwed up.
I have no idea what CL150 is in the power supply, I need a hint.
A few questions linger, you'll spot them. Please reply with answers. 🙂
..Todd
I have no idea what CL150 is in the power supply, I need a hint.
A few questions linger, you'll spot them. Please reply with answers. 🙂
..Todd
The 6GK5 replaces the EF86, not the 12AT7......
.....with a red LED for the cathode R and the 10M45S and related R's on the plate. Look at Poinz' schema for a rough idea of the supporting R's. I'm sure we'll need more guidance from Eli and others on this.......like what's the proper B+ for the 6GK5?
Taj: Very sweet PDF BTW.
Last edited:
And where off to the races! Nice job.
Voltage to OPT center tap should be changed to 465V. At his point, i'd wouldn't mind 2 schematics. One with the EF86 as you have it, but put back the 12AT7, and another with the 6GK5s in place of the Ef86. That last one needs circuit changes to be made. Lets push ahead with the EF86 schematic since it's almost complete, and add the second schematic with 6GK5s later as the circuit changes are figured out. This will give a give choice to builders.
I see you made a change to the splitter CSS. I am interested in the 10M45S used here, but would a Cascode arrangement be best? I know little about the subject of CCS.Looking for simple but effective. The K&K audio type CCS interests me.
Voltage to OPT center tap should be changed to 465V. At his point, i'd wouldn't mind 2 schematics. One with the EF86 as you have it, but put back the 12AT7, and another with the 6GK5s in place of the Ef86. That last one needs circuit changes to be made. Lets push ahead with the EF86 schematic since it's almost complete, and add the second schematic with 6GK5s later as the circuit changes are figured out. This will give a give choice to builders.
I see you made a change to the splitter CSS. I am interested in the 10M45S used here, but would a Cascode arrangement be best? I know little about the subject of CCS.Looking for simple but effective. The K&K audio type CCS interests me.
Last edited: