I would say that the most important audible differences are in that range, but in general the sound is cleaner across the entire spectrum starting in the low midrange, below that there is less difference. (Transformer is 7K primary, plate voltage ~1kV, fixed bias with IT coupling and plate current typically set to 120mA)
Agree completely: the graphites aren't even close to the coppers IMO. I only keep a single pair of graphites just in case.
Cheers,
Kendall
Absolutely, but if you consider that the mean average of music's fundamentals is 640Hz - I'd heard this years ago and have (through spectrum analysis of different music) confirmed it a bunch of times - 700 is a little too close for comfort - I'd cross over lower myself.
Sorry about joining this discussion late, but I've a few things to add (and a couple to dispute). While I too prefer the copper plate, it does have one deficit (and some essential advantages) versus the graphite.
If you are running full-range, the graphite's do give you a more solid bass (once they're broken in) and, in fact, most comparisons I've seen on the internet support this.
"Lampizator", for one - there are others as well.
I will say this, though. The copper-plates are unquestionably better in the mids - (yes, I've used both) - but I find the highs to be comparable between the two (with maybe a slight edge to the copper) and the bass to be better on the graphites. Also, the coppers - for some reason; maybe it was just my pair (I've only had one pair of copper-plates) - seem to break in more quickly. That having been said, once broken in, the graphites do have significantly better bottom end - enough so, that if you were not bi- or tri- amping, you might actually prefer them overall.
Thank Digitrax. There is a question about what horn I actually have (long story) but I just bought the MiniDSP UMIK-1 so I can measure how low the horns go. They were sold to me as 350hz tractrix but they are 1.5" smaller in diamiter and 1" shorter in length than all the other 350hz horns Edgar sells.
I would like to cross at 500hz but measurements will tell if I can or not.
Hi Kevin. I am running 870votls, fixed bias, cap coupled, and 8K trannies. If nothing else it will sound better than my Bose......I hope
Please note I clearly stated "in person" I put relatively less credence in subjective things I read on the internet than when someone I know personally states something to me, and particularly if it conflicts with my own experience.
I am running my amplifiers full range into a moderately complex 3 way Onken based system with a measured F3 of ~37Hz. (Iconic 165-8G woofers, alnico 2440 and 075 JBL horns with OEM diaphragms driven by a Linkwitz-Riley 2nd order XO). Both the graphites and coppers have been compared with several hundred hours of operation on each pair and while not identical I find the bass on the coppers to be quite adequate and not significantly different than the graphites. Transformers are very large heavy Electra-Prints with 7K primaries, and I'm running at 120mA and 1kV.. IT driver stage based on triode connected D3A and operating in constant current mode at 20mA. Just some context for my experience - now about 18 months.. I will probably never use the balance of my graphites. I listen to quite a bit of techno, electronica, industrial, grunge, and similar not to mention some classical where good bass performance will make or break the performance.
Hi Kevin. I am running 870votls, fixed bias, cap coupled, and 8K trannies. If nothing else it will sound better than my Bose......I hope
My sympathies..
I actually still have some of the Bose hardware I helped develop in times long past doing HT duty here, cold the GM70 easily runs circles around that system, warm they aren't even in the same galaxy.. 
grid stopper
Hi Vincent77
I would always install grid stoppers. My personal class A2 833A SE amplifier oscillated wildly the first time I fired it up without grid stopper installed. A mere 22R carbon resistor fixed the problem immediately. I see no reason to omit the grid stoppers. You can use a higher value when running A1 with a cap coupled design.
Kind regards, //WDC
www.monolithmagnetics.com
Hi Vincent77
I would always install grid stoppers. My personal class A2 833A SE amplifier oscillated wildly the first time I fired it up without grid stopper installed. A mere 22R carbon resistor fixed the problem immediately. I see no reason to omit the grid stoppers. You can use a higher value when running A1 with a cap coupled design.
Kind regards, //WDC
www.monolithmagnetics.com
Hi Vincent,
Yes I would recommend grid stopper with the GM70, bear in mind if you ever plan to run A2 that the grid stopper eats up voltage compliance in the driver stage. My current amplifier has 1K grid stoppers on the GM70s and I have not had any issues with stability. I have not tested without so I cannot state categorically that you can't get away without them - I wanted to avoid any possibility of mischief in a circuit with 1kV plate voltage.
Yes I would recommend grid stopper with the GM70, bear in mind if you ever plan to run A2 that the grid stopper eats up voltage compliance in the driver stage. My current amplifier has 1K grid stoppers on the GM70s and I have not had any issues with stability. I have not tested without so I cannot state categorically that you can't get away without them - I wanted to avoid any possibility of mischief in a circuit with 1kV plate voltage.
I don't base how much credence I put in a statement on whether I heard it in person or on the internet, I base it on who's saying it. I would give more credulity to something Lukasz Fikus says on the internet than to a lot of the statements my "self-appointed engineer" friends have made to me personally, particularly if it coincides with my own.
I run both bi-amped and full range depending on whether one of my amps is needed elsewhere. I'm running Electra-Print OPT's as well - (yes, they are "very large" and "heavy") - but at 10K:6Ω (Jack himself will tell you that IMD goes up quickly when you load the GM-70 below 10K). I'm using a choke-loaded 6Н30П-ДР @ 27.5mA quiescent - I didn't have chassis room to build a tube-based CCS and I feel the detriments of a solid-state CCS (e.g. increased IMD from mixing solid state and tube action) outweigh the benefits, but the driver is direct coupled - no IT or coupling cap - and both the GM-70 and driver B+ are tube-rectified. I run 1250 @ 100ma. My speakers are 3db down anywhere from 30-36Hz, depending on where I have the bass traps positioned in the room - (unfortunately the best bass extension is also the worst living arrangement, and creates a "hump" at 90-100Hz, so I usually have to settle for an F3 of about 33Hz).
I'm not disagreeing with you, Kevin - I prefer the copper plates 100% of the time when I'm bi-amping and 90% of the time when I'm running full-range, but bass is one area where the graphites win out - (Wp is another). I don't listen to techno or industrial grunge and any music "where good bass performance will make or break the performance", but I can hear the difference on pipe organ, upright bass, and tympani - (put on the Cozart/Fine "RCA Living Stereo" recording of Saint-Saëns Symphony #3 and tell me you don't hear a noticeable difference between the copper and the graphites in bass).
I would at least allow that some people might prefer the graphites.
I don't, and you don't, but some people might.
Last edited:
Benefit: Transparency/less in signal path
- Good interstage trafo's are expensive, and they still add distortion and cause phase anomalies. Caps are often worse.
These things only exist because of the mismatch between the voltage coming out of one stage and the voltage you'd like the grid of the next stage to be at.
By using a separate power supply for your driver stage, below the ground of your power stage, you can set the voltage coming out of your driver equal to the voltage you'd like your power tube's grid to be at (and use nothing but a wire between the two!)
Downside: Power supply complexity and/or cost
- Let's say your power tube's cathode is 10V above ground and you want to bias the power tube at -120V. If the driver puts out 160V quiescent, the ground of its power supply has to be 270V below the power tube's ground (-270V + 160V -10V = -120V). This power supply has to have it's own power transformer (or power transformer secondary), its own rectifiers, and its own CLC filter and/or regulator circuit, but... your driver tube provides both signal and bias voltage to your power tube, eliminating the need for a bias resistor, coupling cap, or IT.
There are lots of "DC coupled" or "direct coupled" amps on the web you can study, but unfortunately, not too many for the GM-70 - probably because of it's high anode voltage; the idea of stacking another 200-400V underneath what's already 1000-1500V probably doesn't appeal to a lot of people!
Hope I've explained that well enough. Feel free to ask more if I haven't.
@digitrax, great discussion, and thanks for sharing some additional thoughts. Interesting comment on credibility, I generally listen to a very select few who I regard as being particularly credible, and even then I don't always agree with them..
In the case of my particular design the CCS in the driver stage sets the plate voltage required for 20mA of quiescent current in the driver stage, and actually performs no AC function other than to provide pretty good PSRR - local capacitor decoupling sets the AC source impedance of the supply feeding the IT.
I'm actually not a big fan of soiled state CCS loading despite the many theoretical advantages stated by their supporters. (In fact I am not a fan of SS in tube signal paths in general, but I'm probably irrational on that subject. )
Interesting comment on IMD performance, I choose 7K:8 deliberately in order to reduce distortion at the expense of power. I've not measured IMD in this amplifier but at some point should be able to do a multi-tone FFT based measurement and see what comes out of that. (Not imminent due to lack of time) Some people run them as low as 5K..
In the case of my particular design the CCS in the driver stage sets the plate voltage required for 20mA of quiescent current in the driver stage, and actually performs no AC function other than to provide pretty good PSRR - local capacitor decoupling sets the AC source impedance of the supply feeding the IT.
I'm actually not a big fan of soiled state CCS loading despite the many theoretical advantages stated by their supporters. (In fact I am not a fan of SS in tube signal paths in general, but I'm probably irrational on that subject.
)Interesting comment on IMD performance, I choose 7K:8 deliberately in order to reduce distortion at the expense of power. I've not measured IMD in this amplifier but at some point should be able to do a multi-tone FFT based measurement and see what comes out of that. (Not imminent due to lack of time) Some people run them as low as 5K..
I think we all do, and Lucasz is one my select few, so it's hard for me to regard his opinion as just "something on the internet", hence, my leap to his defense.
I don't think you're irrational at all. There is a goodf deal of evidence to support the fact that (at least in most instances), the combination of sand and tube action results in higher IMD. AFA "the many theoretical advantages stated by their supporters": size, efficiency and cost; you can do anything you can do with a solid state device in tube form, but it may require more power, take up more space and cost more, (but it will sound better).
I don't even like the idea of using SS diodes to get DC for filaments, but it's a necessary evil.
Yeah, you can get "useable" results down to 5K, but those who've experimented report that you start to really hear a difference once you get up around 7-8K. Remember, although the GM-70 has a lsightly lower Rp than the 845, it is actually more sensitive to loading than the 845. Fortunately, it also takes higher anode too, so you can actually load it at 10K, drop 1250V across it (/10,000=100mA x 1250=125W or full dissipation) and get 25W in pure Class A; (alright, a little less for copper, but that's without driving into A2, which you should be able to do if you've got enough current on the driver). My first OPT was an Edcor 8K:4/8/16 and it sounded pretty good. Then, I got them to make one at 10K:8 and I could hear some improvement, particularly in the bass, but also in imaging and lower-mids. Things really opened up when I got the Electra-Print 10K:6Ω with 100mA (25Hz F3) core - (I should receive today my 4th trafo: same as the other Electra-Print, but with 120mA core; I'll let you know how it compares) in terms of depth of field and definition and detail. Still, I was surprised at how well the Edcor performed (for the money) -- better than the more expensive Hammond I borrowed from a friend. I'd really like to try Per Lundahl's OPT's for a GM-70 amp, but everything in that impedance range is rated 70mA saturation, and a custom job, I think, would be pricey.
As you may have noticed, I'm a big fan of keeping IMD as low as possible. Think about it: if even odd-order distortion is objectionable because it's not an even multiple of the fundamental, imagine how awful a frequency which is not a multiple of the fundamental at all sounds! IMD is just downright ugly and even a couple of points make a big difference - much more than THD. If there's an upload/file section in this forum, maybe I (or someone else) could post some standard SMPTE 60Hz/7KHz 4:1 test files (as .wav's or .aiff's), so people can shove them through their circuits and into RightMark (or Clio or whatever). I just upgraded my little 100W capacity jig (that puts an 8Ω load across the amp and knocks the output down to line level for scope or computer input) by replacing the 1% metal films (on the "line out" side) with Vishay 0.01% metal foils so I'm going a little measurement crazy these days.
Well, thanks for entertaining my verbose rants. Cheers.
Hi.
I have ordered a pair of Lundahl LL1688 for my GM70 amp (not yet finished), custom gapped for 90mA at no extra cost. They can be wired for either 5K5 or 9K2 impedance.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.