Joplin Builder's Log for KYW with pictures
1/30
Hi,
I have the Joplin here too and observed the same. Also, I feel that the adjustment
process is a little involved and very "jumpy", not what I'd recommend
for first timers.
I have been discussing with Brian several mods to the Kit. If Brian could confirm the wire current rating for the Bias winding on the mains transformer and the core rating (Watts) I could see the Circuit with several modifications to allow different Operation for different people.
1) Self Bias Class AB1 2A3 - 375V +B with 80mA per pair quiescent current and 12 - 15 Watt power output
2) Self Bias Class A1 2A3 - 300V +B with 120mA per pair quiescent current and around 7 - 9 Watt power output
3) Sef Bias Class AB1 45 - 365V +B with 65mA per pair quiescent current and around 6 - 7 Watt power output
4) Self Bias Class A1 45 - 300V +B with 72mA per pair quiescent current and around 3.5 - 4.5 Watt power output
I can also see an option to use 300B's in the Amp as there are by appearances four seperate 2.5V/2.5A heater windings, so a 300B Push-Pull Amp at a very low voltage operating points should also be on the cards. Maybe even one at 375V +B & 120mA per pair of valves in Class A1.
The real question is if the mains transformer has enough reserve power... Or maybe if an upgraded Powertansformer may be on offer at some time?
Anyway, let's see how this goes.
Sayonara
2/12
Hi folks,
The last few days I have been working on a Joplin Amp that ended up with me
already factory build, for review. Having put the Valves in and set the Amplifier
up for use I found the neccesary process of adjustment both rather twichy and
longwinded (bias & hum). In addition the HT settled at around 330V when
set up for Class AB1 with 2A3's which made clear that the quite ineresting use
of the Amplifier with 45's would not be possible.
After some discussions with Brian we agreed to apply an "officialy approved" Class A, selfbias modification to the Amplifer. In the context I must especially thank Lance for his work on the PSU and his mods etc.
In addition, Lance's work pointed out that at low frequencies the Amplifiers output trnasformer lacks enough iron to support large power outputs, so the reduction output power from the 11...13 Watt at 1KHz for class AB1 to around 9 - 10 Watt in Class A1 (actually, the last watt or two are still class AB1 anyway) is in reality pretty inconsequential.
In my modifications I considered three key factors:
1) Simplicity
2) Comercial Impact (how much change would be needed to the kits)
3) Increased flexibility
I made provisions for the Amplifer to be used with either 2A3's or 45's or 300B's. My own modified unit is switchable 2A3/45 (different cathode resistor) and can be converted to use 300B's by changing two wire bridges on a Tagstrip (per channel).
Having looked, standard 300B's (WE, Svetlana, Sovtek, Valve Art) mechanically fit the Joplin fine, bigger bulb 300B's (and incidentally 2A3's) will not fit without touching glass. Brian suggesed that the final version of the Joplin will have a wider spacing of the Output Valve Sockets, which should allow the TJ Globe and AVVT Valves to fit.
In order to simplify assembly of the self bias arrangement I fitted a tagstrip on the Valve sockets to terminate the heater wires and hold the self bias resistors. The heater wires are terminated on the tagstrip so that easily can be connected in series (5V/2.5A for 300B Output Valves) or Parallel (2.5V/5A for 2A3/45). I also connected the black groundwire there and the two cathode resistors each bridge across one of the sockets to a tagstrip on the other side of the socket.
I used only ONE common cathode resistor for both Valves and only ONE Humbalancing pot (which is needed anyway to allow the use of 5V series heaters. That means that if some slight hum is acceptable the Cathode bypass capacitor can be completely omitted. Otherwise the AC signal applied to the Cathode bypass capacitor is only the difference in AC current between the halves, which significantly reduces the signal on the cathode bypass cap and thus the influence on the sound.
If seperate cathode bypass capacitors and resistors are used the signal current loop is always closed through the two cathode bypass caacitors, clearly making the qulaity of these capacitors more relevant.
In order to use my favourite, NOS, mintgreen, noninductive, endcapless construction Painton power resistors in the cathode and in order to to make the cathode resistor switchable for 45's I used a 680R (12W) in series with 56R as permanently connected cathode resistor.
This gives with the approximatly 12R effective DCR of the Hum pot a 748R Bias Resitor for the 45's, spot on for a pair of 45's with 250V Anode Voltage according to the RC-10. A further 820R (12W) resistor is switched in parallel to the 680R + 56R for use with 2A3's, for a 400R effective Bias Resistor.
Without cathode bypass capacitors I was unable to get the the hum on the output below around 3mV. With low sensitivity speakers (90db/2.83V/m or lower) that may be tolerable, in my system (effectively 97db/2.83V/m) 3mV are not tolerable. The way around this would be either to use DC heaters or to use a cathode bypass Capacitor. I used a pair of Nichicon GZ Series (low impedance, super long life) 330uF/250V Capacitors plus bypasses for this.
With cathode bypass capacitors and a single Hum balancing poit and common cathode resistor the noise in one channel is 0.6mV, in the other it's 0.4mV. With respect to the nominal 8 Watt RMS into 8 ohm this translates into 82.5db and 86db UNWEIGHTED Signal to noise ratio with the stock Sovtek 2A3's, not bad at all for an AC heated Amplifier without negative feedback.
Having now self bias in place I changed the gridleak resistors over to 330k in order to avoid the original amplifiers rather early LF cutoff (100k/0.1uF = 16Hz -3db). This will also allow the use of smaller value coupling Capacitor, the ones I have in mind are 0.047uF ACL Silver Mica types of course (not cheap but by far more worth the long green layout on them than any of the "boutique" Cap's)... For now I chucked the Wima MKP's out and replaced them with ERO KP1832 0.075uF/800V Foil & Film capacitors (from www.steinmusic.de). When doing this note that the 0.47uF Capacitor in the phasesplitter circut (grid of the second 5687) also has a big influence on the sound. The 0.47uF Value is excessive, anything from 0.047uF upwards will be fine, quality over quantity is the motto there as well.
The Joplin I have contained a pot below the chassis per channel that seemed to be intended for either bias balance or AC balance (if intended to balance the bias between valves they do not work BTW).
By keeping this in place and connecting the wiper to ground, instead of connecting it the bias I removed the fixed Bias. By adding a pair of 1M resistors from the two sides of this pot to the +B supply for the input Valve I allowed the slightest degree (around 2V when centered) of positive grid bias, which can be used to balance out the differing currents between non-matched output valves.
I do consider this application to be optional, as said the stock Joplin does not allow this balancing, so normally I would leave the pot and resistors completeley off and simply ground the lower end of the gridleak resistors. The two pairs of 2A3's I received with the Joplin showed a near perfect match (less than 1mA difference in current) for one pair, while the other pair had +/-4mA deviation between the two Valves at nominal 40mA, meaning that I ended up with one Valve set to 36mA current and the other to 44mA. If the manufacturer accepts such a mismatch without attempting to allow correction clearly there is no reason for me to do otherwise.
I only included the Balancing feature for using potentially severely unbalanced "Hamfest find" 45's or the like, wher emaking up two pairs of mechanically and manufacturingwise matching Valves can be a challenge.
In a chase for some low level hum I found that frontened heaters on the Joplin where centertapped and grounded. I have repeatedly seen this in Opera/DIY Hifisupply gear and in each case I have been highlighting this. Doing so will leave the parasitic heater/cathode diode to conduct and to inject all sorts of (low level) noise into the circuit.
As I was no longer using the fixed bias components sections I used this area to fit 3pcs of 47k/2W resistors and a 47uF/450V Capacitor to bias the frontend heaters up to around +90V. Taking the centertap of the heaters from the ground on this PCB to the +90V sections killed the hum. From experience I also know that this heater bias tends to improve the overall sound. I would recommend to implement it in the final Joplin kit using 100k/39k/47uF with a slightly rerigged PSU PCB.
I applied a final few mods in the Frontend simply to shift the input stage operation more to a "Kondo" sound and to reduce overall the Amplifier gain by 12db (no Negative Feedback) so that full output requires around 500mV, not slightly over 0.1V. These mods I will cover if there is interest, they are more a personal thing and have nothing to do with the Class A mod.
Anyway, per channel my suggested Mod requires:
1 pcs standard tagstrip (will broken
apart to make three sections)
1 pcs Resistor 390R/12W or 2 Resistors 750R/7W
1 pcs 220uF or larger Value 150/160V Electrlytic Capacitor
2 pcs 330k..470k 0.5W Resistors (Gridleak)
Process:
Remove one of the heater hum balance pots, optionally install a 2-Pole switch in it's place to either select heater voltage or cathode bias resistor (switch will switch either 2A3/300B or 2A3/45) or a bias balance pot.
Install Tagstrip on the 2A3 Sockets using the original socket fixing screws.
Rewiring the heaters of the sockets in parallel and provison of easy reconnection of heaters for 5V. If a heater voltage switch (2A3/300B) is desired wire up switch, otherwise solder in wire bridges.
Install either a 390R 12W+ Cathode resistor (use only with 2A3/300B) or a pair of 750R 7W+ on the tagstrip, plus wire if desired the switch (switchable for 2A3/45) disconnecting one of the resistors on demand.
Install (via sticky pad and tiewrap) one each cathode decoupling capacitor to the side of the chassis (optional - only needed for lowest noise).
If desired (re)wire the AC balance Pot for bias balancing with two additional 1M resistors per channel. If bias balancing is considered VERY important the Pot may be installed instead of the removed heater humbucker pot, in which case really some external testpoints should also be provided. I simply adjust for a minimum DC Voltage difference on the Anodes. Under 10mV is easily done, in which the imbalance is way below 1mA.
Anyway, with the modded Ampliifer you can operate 45, 2A3 and 300B Output Valves with minimal changes (I suspect a suitable 4-Pole / 3-Position rotary switch could be fitted instead of the toggle I used to switch 45/2A3/300B without opening the Amp up). The Driverstages also allow a range of Valves from the 5687 and similar range, including 6900/7119/E182CC while the input Stage Valve can be a ECC81/12AT7/6021, a 12AV7/12AZ7 or a 5965 or a 12AY7/6072A or even a 13D3 or a ECC82/12AU7/E80CC. This should allow for a wide range or "tuberolloing" and "soundshaping".
All in all I think the Joplin Kit is a nice little Amplifier "on the side" which is very versitaile with much scope for fun. It comes highly recommended for these reasons.
As the batteries in my digicam needed recharging I'll post pictures of the mods in the evening when they have recharged. I will provide Brian Cherry with a schematic of both the Class A mod and the frontend Mod which he will no doubt be happy to send out.
Short listening to the modded Amp showed the sound to be considerably more open and natural than stock and lower noise to boot, I could not try out "max power" as it way past midnight. More impressions later. On the bench power measured slightly higher than Lance measured, a result of the slightly higher voltage UK Mains.
Sayonara
2/12
Hi all,
After playing a little more with different Valves on the testbench I must note
that is lowest possible noise is of the essence you MUST have the have the Bias
Balance Pot if you do not use tightly matched output Valve Pairs.
With around 3-4mA imbalance in each channel from the amplifier being bias compensated for the 45's (which are used and vary a bit) and using the Sovteks 2A3's I ended up with being able to reduce the hum only to about 1mV (0.8mV one channel, 1mV the other). Balancing the bias with the 2A3's gives nearly 6db lower noise (0.4/0.6mV respectively).
On the other hand with the Bias Balance Pot set to Center (as it it was not present) the maximum Difference between all four Sovtek 2A3's was under 1mA and the noise was minimised, no sweat. So, if you use new production matched Valves omit the Bias Balance, if you want to be able to handle "all commers" fit it to the top of the Amp and drill an extra hole for the 45/2A3/300B rotary switch and a pair of testpoint jacks (unless you fancy taking the bottom off everythime you change output Valves. I am likely to do exactly that (Bias Balance on top plus big rotary switch) in the near future...
Not much more listening yet, I'm letting the Amp's burn in....
Here some snaps from the inside of the Joplin's playing here, for those interested. A general overview of the chassis....
In the PSU Section note the added Snubber and homemade "bridge" rectifier using 8pcs Motorola MUR460 Soft Switching, fast recovery rectifiers. My snubbers are 2 X 12 Ohm and 2X 0.33uF, the capacitos fitted in the posiions originally occupied by the diodes feeding the negative side (ground) of the rectifier.
A look at the lefthand side of the Valve Sockets show the added Tagstrips and resistors, plus the bypass Capacitors (Evox/Rifa 22uF/63V MKT & 0.05uF/400V Micamold) for the Cathode bypass capacitor.
As can be seen, there are still miles clearance for the Volume Control. The capacitor mounted on the side is the actual cathode bypass capacitor.
The righthand side of the Sockets looks mirrored, exceting the indicator light emitting diode replacing the Pot.
On this photo it is a little easier to tell how the heater arrangement is done. From the five tags in the middle of the upper long tagstrip the center one is ground and the two outmost ones are one wire each of the two 2.5V heater windings of this side, notably the opposite polarity ones. The two remaining tags have respectively the other wires, but in this case the wire next to the outer one is from the OTHER winding.
Thus bridging (as seen) the two outer pairs of tags will give paralleled 2.5V heaters. Bridging the two inner pins with each other on the other hand gives 5V.
Circuit diagrams both for the "Output Stage only" mod and the slight frontend changes found in my unit make their way to Brian and will not be posted here by me (if Brian wishes to post them - fine with me)
A few more shots of the Amplifiers....
Here a closeup of the trio of input Valves (my own ones, not stock)
A pair of Cunningham 45's up close and personal (I just wish they could mesh anode globe 45's)
(Scott) Joplin boogeeing it down
with 45's... (4 X 45 Caliber - does this qualify as Colt Cloverleaf?)
Its a Sovteks life... (Blunt Soviet offensive Valves conquer Great Britain - if only Churchill knew...)
Ain't misbehaing - Joplin all caged up (GAWD is this cage UGLY, fallen right out of the BIG UGLY tree and hit all branches on the way down...)
Sayonara
2/16
Hi all,
Today I applied myself to an extended "tube rolling" session. This
included a repeat of Sovtek 2A3, Sovtek 2A3 as Pseudo 45 and real 45's, as well
as various input and driver valves.
First, as previously real 45's are "the real thing", yet the Sovtek (single anode) 2A3's have a surprisingly close tonality, only giving up a little finesse and detail to the 45. Swithcing teh Sovteks to a 2A3 Operating point made the sound brighter, harder and less pleasant, more 2A3 like in essence.
Nor for the Input & Drivers. I had started with Westinghouse 5687's and JAN 6072A. We first tried various input Valves, all of which can be used without severely affecting the function of the Amp, though the Driver Valve operating point will shift slightly.
Short and sweet, the GE 5-Star 6072A came out tops, followed by the RCA 12AY7 and GE Cleartop 5695, then the JAN 6072A. All those Valves had a very similar sound, with the tow best resolving snares and brushes much better, the 5695 clearly had the most bass impact, at a loss of finesse. We also tried various ECC81 and ECC82 which by comparison sounded merely competent, nothing special.
In a comparison between the Westinghouse 5687 and the GE 5-Star 5687 I ahve around the GE again had the edge. proving more liquid and natural sounding, very notable on the Sax.
So my recommendation for the Joplin is to run 45's in Class A, with GE ***** 6072A and 5687 as input/driver valves. Not neccesarily the cheapest option, but IMHO very well worth it.
Compared to my "Lady Day" modded Billies the Joplin is more of a cerebral experience, lacking some of the tonal colours and sheer involvement, as well as the final degree of dynamics (even with 2A3's fitted and run as 2A3's) but overall this amplifier is quite good, free of graininess and obvious Push-Pull "sound". In some ways it reminds me of the Billies with AVVT 300B Outputs instead of TJ's.
It is a very valid alternative to other options and very well worth considering for "DHT" sound, with a little less bloom, warmth and intimacy than classic SE Sound. The way the amplifier plays now I feel no urge to go back to the "Lady Day" Billies, though no doubt in the end I will do so.
I have also just stuck a deal on probably three pairs of Sylvania 10Y's, which should give me the chance to try these in the Joplin, probably later this week and hopefully before the ETM fromal review.
Anyway, for those who want an authorative answer as to the question:
"Which Amp is better, Billies or Joplin" the answer is "neither" or "depends". Both amplifiers sound good, are low noise and give "DHT valve sound". They also sound considerably different, enough so that the "either or" answer depends strongly on the system context. My personal consideration would be that it is very well worth having both Amplifers, especially if you own a lot of DHT's, including 45's, 2A3's, 10's and 300B's, as some of us do...
I suspect the stock Billies will go subjectively as loud on a given speaker as the Joplin, with a gridchoke fitted probably a little louder.
Ciao T
PS, it is noteworthy that the phasesplitter in the Joplin is pretty well balanced where impedances are concerend, but is unbalanced with regards to voltage, giving a harmonic spectrum very similar to an SE Amplifier....
But is it Still a Joplin?:
Sept 25, 2003 Joplin Gets a Heart Transplant
Here's the latest and greatest from KYW:
