Indeed G2 tweaks and bias adjustment

[Deleted]

The metal casing does have enough surface in itself BUT because of it's thickness and the holes also a very high thermal resistance.
This would result in a hot spot where the parts are mounted and the heat not flowing through the metal (high thermal resistance) so not enough 'effective' cooling area alas.

You could mount some regular heatsinks (black) inside the casing.
There seems to be enough room for that.

There is no cost effective way to make a linear PSU I am afraid (unless you can salvage the parts somewhere or in some dump store).
You will need a transformer of about 30W (toroid), a firm capacitor and cooling for the linear voltage regulator.
All costing money when bought new.
That is why SMPS are 'great' for higher currents... very cost effictive, no expensive copper and metal, no big cooling fins, no big capacitors, small housing, cheap parts as they are made in huge quantities... e.t.c.

[Deleted]

BUT because of it's thickness and the holes.... This would result in a hot spot where the parts are mounted

That's what I suspected, hence asking.

Thanks Frans.

[lynxo323]

Hi Everyone,

just got my G2.

have few questions. thanks for answering.

1. what's your favorite tube for the G2? 6V or 12V tubes?

2. when biasing 12V tube, do you bias to 13V as well?

3. What is the PN# for the Gold Pin Neutrik heaphone jack?

4. I have the ECC82 JJ Gold Pin. It's really thick and bass heavy.
What should it be biased too?

5. Does Linear 24V supply make any difference?

6. so consensus is to leave 1000uF 85C caps alone? or change to 2200uF 85C or 105C?

[lynxo323]

Hi,

hoping someone will reply.

How does the G2 once modded compare to other tube HPA's?

Does a higher voltage(>100V) 12au7 sound signficantly better than this 24V version?

I'm also pondering the 12FM6 Tube MilletMax. For the G2 circuit with the 12FM6 Tube, would it sound much better?

I'm new to tube HPA's so appreciate your replies.

Thanks

[Deleted]

It's a very 'warm' listen in comparison to some other tube amplifiers. Not as extended in the treble and Frans found that there was a lot of cross-talk. I use a 6922 on mine at the moment and save the nice tubes for another occasion.

It's powerful with a heck of a lot of gain and easily drives Senns and even the K701. For me, it warms the K701 up a bit. It can get a bit too hot and is prone to interference.

With all of these quirks, I still like the sound of the amp. It has quite large 'faults' and also there may be quality control issues but the basic sound is rich and involving.

I think that whether you like it or not may depend on what you yourself prefer in terms of sound. Some listeners prefer a treble focussed sound and others prefer a warmer sound. This amp provides a warm sound that some people do not like.

My favourite tube amps are the X-Cans. Really good value for money and way ahead of the G2, although the G2 is very cheap.

One surprise for me was also the V-Can which, once it has been mooded, almost sounds like a tube amp, without the associated heat build up. Absolutely superb and a bargain at £60 (Plus the cost of modding of course) - but at least you can then upgrade it.

[Deleted]

If you have IRF type MOSFET's bias the amp around 15.5 Volts.
If you have IRL type fitted bias around 13.5 Volts.

The few types of tubes I have rolled in there (3 types not yet 12V) all had exactly the same flat frequency response.
The lower limit is solely determined by the output caps/output load and the upper limit mostly by the MOSFET (when cathode caps are used)

Higher voltages on tubes makes them amplify more linear with less distortion.
this distortion is the 'charme' of these low voltages hybrids.
Xcan (as far as I know) has double the power supply voltage then these amps and lower anode voltages then 'normal' tube amps.

Although the G2 has a switch between 12V and 6V types the 12v types do receive a 20% too high glowing voltage when used.
You can correct this by altering the LM317 bias resistors or a another possible correction for this can be found in the pdf file.

The 1000uF output caps could be replaced by better type 1000uF/25V 105oC or 2200uF/25 types (105oC) watch its physical height for this value !

[lynxo323]

Thank you Rabbit and Solderdude for your very helpful replies.

Thes G2 is my first retail tube hpa. I actually built one p2p from diyaudioprojects, the low voltage 12au7 headphone amp.

I tried the stock circuit at 12V, later tried +18 for the +B and for heater used LM1084(heater for 12.3V). both were with the ECC82 JJ gold pin tube. I guess it's related to this JJ tube, since in all cases, including the G2, it sounds very thick and bass heavy.

I have the latest G2 version that uses IRF510's. eventually I will do some mods. The 6N23P included with the G2 sounds okay.

I will have to try a higher voltage +100V version then to see how much improvement from these low voltage types.

in the mean time, I would like to try tube rolling. I'll give 6922 a try. which is a good one? how about 6DJ8? does 6CG7 have enough gain or it's no good at low voltages?

[lynxo323]

quick question. what is the characteristic of the left/right channel biasing?

why when you tweak one channel, increase the bias voltage,
other channel voltage goes down?

thx

[Deleted]

In principle left and right bias voltage are independent.
The bias of the other channel is affected if the bias voltage was set too low (as the manufacturer does around 12V)
Above 14 V with IRF this should not be the case anymore.
This is why I recommend/use IRL510/IRL530 because these do not have this 'affliction' above 12V.
12V tubes (they also run at 6V but in parallel) have this effect till a higher anode voltage then 6V types.
This is due to the fact that the heater receives a heater current (both channels class-A currents minus the current though the 2 big resistors) and not a fixed heater voltage.
It is the same reason why the crosstalk occurs at lower volumes when the bias is set too low.

Reason: When you turn up the bias voltage the glowing voltage rises too because the LM317 finally comes into proper working range causing the tube to change it's emission and therefore anode voltage too.
Above a certain voltage the heater voltage doesn't increase anymore and this phenomon stops.

Another reason to use the separate heater power supply as explained in the pdf.

Optimal bias voltage with IRF = around 15-16V
Optimal bias voltage with IRL = around 13-14V

NOTE: At these higher anode voltage settings the LM317's get very hot !

[lynxo323]

excellent explanation.

for ECC82 tube, this can be used at 6V or 12V.

so it should sound better with slider switch to 12V for heater?

yes, I would like to replace IRF to IRL but look like requires further modding with larger LM317 heatsinks.

btw, after G2 has complete mods, how does it sound vs. other tube HPA? is it still a so so tube hpa or very high quality sound?

there is a Triad AC adataptor 24V at 1.2A (38W). This linear psu should be better than the switching one. is 1.2A enough?

thx

[Deleted]

ECC81, 82, 83 = 12V
Both filaments will be receiving about 7V (so a ECC82 12.6V tube would be on 14V effectively)

I would recommend bigger heatsinks for LM317 anyway (or black types) but you can use IRL with the current heatsink.
Set the bias around 13.5V in this case (6922 or ECC88 tubes).
When using 12V tubes a 14 Volt bias is better.

The LM317's will be equally hot as when you use IRF and bias around 15-16V.

After all the mods the sound is better.
Distortion is lower.
Asymmetric distortion at higher output levels are completely gone as is the crosstalk.
tube swapping is easier.
Setting bias is much more stable/easy.

I think the Triad DC wallwart is NOT regulated so wouldn't use it.
The 25V input cap would probably receive overvoltage as non regulated DC PS only have the specified output voltage at the specified current.
At lower currents the output voltage is higher.
Also a slight hum might be audible as the amp is very sensitive to DC variations.

Only regulated 24V DC power supplies can be used.

If there are sonic benefits when using regulated DC power supplies I do not think so.
There might be if the SMPS has high common mode RF signals and other components in the audiochain also have switching components and these interfere there MIGHT be annoying by products audible.

I would only go for regulated lin PS if the original one stopped working or annoying whispering noises or tones are audible when connecting the audio input of the amp to other equipment.

[Deleted]

If there are sonic benefits when using regulated DC power supplies I do not think so.There might be if the SMPS has high common mode RF signals and other components in the audiochain also have switching components and these interfere there MIGHT be annoying by products audible.I would only go for regulated lin PS if the original one stopped working or annoying whispering noises or tones are audible when connecting the audio input of the amp to other equipment.

Frans
With other HAs such as some from Graham Slee, there have been improvements when the SMPS was replaced by a linear PSU.
The problem with SMPS is that coupling capacitor (usually around 100nF) between Primary and Secondary sides of the SMPS. This cause a readily measurable high impedance noisy A.C. component, which can exceed 100VAC with a 230VAC supply when measured to earth from the ëarth"side of the component with a DVM. It can also be felt in many cses if you rub a finger across the metalwork of the unit.Earthing the case can also result in an audible improvement in some cases. It is sometimes referred to as like "lifting a veil".
Alex

[Deleted]

If the open voltage that can be measured is higher then half the mains voltage turning the wall plug 180 degrees helps too.

a little sidetrack from this thread ...
(Not meant for techs as they already know this but for those that like to have Alex's justified remark explained further)

This 'leakage' is also true (in a much lesser extend) for linear power supplies.
The leakage currents are much lower of coarse then those from the EMI/RF coupling caps in the SMPS but are still there.
the idea behind these caps is to carry away HF/RF common mode signals on the output of the SMPS to the mains in order to comply with rules.
Negative side effect is the leakage of 50/60Hz components through these caps as long as they stay within the norm for maximum current (forgot the exact value but thought they were not allowed to exceed 20mA)

In linear PS's the leakage currents are caused by the primary and secondary windings being close to one another and thus forming a leakage capacitor.
Mostly though it is VERY dependant on the way the L and N are connected due to the way the transformer is wound and the way the coils in the transformer are 'separated'.

Actually I always do the 'humming finger check' with EACH separate audio component.
Important NOTE: the device must NOT be connected with anything but the mains so all other wires have to be unplugged.
Do it once with the mains plug in one direction and once with the mains plug plugged in 180 degrees turned (upside down).
Not possible in some countries though.
This trick WON't work with apparatus that has the safety ground connected (3 pin plugs) into a fully connected (grounded) 3 pin wall socket.

Another way is to use a (high ohmic) DVM (Digital Volt Meter) in the AC volt position with one lead connected to a safety ground and the other touching the casing or preferably one of the ground shrouds of an RCA plug or those cheap LCD based L-N search screwdrivers as well, in this case you'll only have to hold this screwdriver and notice the voltage level when you touch the ground/casing.

When all equipment is set for lowest leakage current connect them all together and you are likely to have the best possible results with this system.

Wether or not a lin PS will give a noticeable performance improvement depends on what it is the amp is connected to, the topology of the SMPS (wether it has a, connected, safety ground or not) it replaces and if somewhere in the chain there are more SMPS or leaking lin PS's and if somewhere in the chain a safety ground is connected.

a lin PS DOES have more benefits in this department then SMPS, that is for certain, so in this light it is always better to use a lin suppy as I have always adviced in the past.

[lynxo323]

After all the mods the sound is better.
Distortion is lower.
Asymmetric distortion at higher output levels are completely gone as is the crosstalk.
tube swapping is easier.
Setting bias is much more stable/easy.

thank you.

I will do full mods then. I was trying to determine if it was worth the extra time/money, or saving those dollars for an x-can's or something of that sort.

I tried ECC82 with heater switch to 6V. One side of the tube heater does not glow and no audio of course. why would it not work at 6V?

[Deleted]

The G2 is no X-Can for sure. It's a good listen but the X-Can is better imo.

[Deleted]

The ECC82 (81,83) have 1 filament end (+6.3V) on pin 4, 1 filament end on pin 5 +(6.3V) and the common pin = pin 9 (0V).

When these tubes are operated on 12V this voltage needs to be between pin 4 and 5 (6.3 + 6.3 = 12.6V).
When these tubes are operated on 6V pin 4 and 5 must be connected (both 6.3V) and pin 9 is 'ground' (0V)

When the switch is in the 12V position pin 4 and 5 are connected so both heaters will receive 6.3V with respect to pin 9 (0V).
When the switch is in the 6V position pin 5 is connected to ground (as well as pin 9) so this heater receives NO voltage, the other heater does so only 1 heater works in the 6V setting when a 12V tube is used.
Beware that this working heater receives an overvoltage !

Be carefull with the 6-12V switch and set it to the correct position before switching on !

[rowuk]

I am going to build one of the 12AU7/ECC82/IRL510 amps using point to point wiring on veroboard. I am also thinking about using higher voltage on the tube. Is the recommended 250V the best?

[lynxo323]

The ECC82 (81,83) have 1 filament end (+6.3V) on pin 4, 1 filament end on pin 5 +(6.3V) and the common pin = pin 9 (0V).

When these tubes are operated on 12V this voltage needs to be between pin 4 and 5 (6.3 + 6.3 = 12.6V).
When these tubes are operated on 6V pin 4 and 5 must be connected (both 6.3V) and pin 9 is 'ground' (0V)

When the switch is in the 12V position pin 4 and 5 are connected so both heaters will receive 6.3V with respect to pin 9 (0V).
When the switch is in the 6V position pin 5 is connected to ground (as well as pin 9) so this heater receives NO voltage, the other heater does so only 1 heater works in the 6V setting when a 12V tube is used.
Beware that this working heater receives an overvoltage !

Be carefull with the 6-12V switch and set it to the correct position before switching on !

with the ECC82 and slider switch to 6V. I get NO glow in heater on one side. it's dead.

if I switch to 12V, heater glow in both side of tubes. so is it tube or G2 issue when slider switch is set on 6V?

thx

[Deleted]

@ Rowuk,

I don't recommend to go much higher then 48V (capacitors rating will have to be 50-63V and cooling will have to dissipated a LOT more heat).
The class A current needed to drive low-ohmic headphones at that output levels would be around 300mA meaning 7W heat from IRL/IRF and 6W from LM317 !
VERRRY HOTTTT.



ECC82 = 12V tube and both heaters will ONLY work in '12V' position.
ECC82 on 6V setting will only light 1 heater, the other heater is shorted by the switch and receives nothing.
The tube and amp are NOT defective.
It's working as it should.

in the 12V position both filaments get 7V (so actually 14V when in '12V mode')

So use ECC81, ECC82, ECC83 ONLY in 12V position as they are 12V tubes.

[lynxo323]

ECC82 = 12V tube and both heaters will ONLY work in '12V' position.
ECC82 on 6V setting will only light 1 heater, the other heater is shorted by the switch and receives nothing.
The tube and amp are NOT defective.
It's working as it should.

in the 12V position both filaments get 7V (so actually 14V when in '12V mode')

So use ECC81, ECC82, ECC83 ONLY in 12V position as they are 12V tubes.

got it! Thanks!

[lynxo323]

Hi Solderdude,

Is 6V rating for cathode caps suffice?

Thanks

[lynxo323]

Hi Solderdude,

can I get away with using this heatsink for the LM317?

www.aavidthermalloy.com/cgi-bin/stdisp.pl?Pnum=581202b02500g

can I stick with stock heatsinks for IRL510?

I understand it will get very hot. I would like to avoid mounting
parts off board. I understand parts will not last as long, which is fine.

Thanks

[lynxo323]

Hi Solderdude,

I did most of the mods/upgrades.

1. 3300uF 35V Panasonic FC
2. 2200uF 25V Panasonic FC
3. Wima 0.1uF 250V
4. 30R output resistor to 68R
5. CCS Resistor to 5.6R
6. IRL510 Seperate TO-220 Heatsink
7. LM317 Seperate TO-220 Heatsink
8. Remove Input switch/3.5mm jack
9. Replace Trimmers with sealed ones
10. 470uF 6V oscons

Crappy On/Off Switch broke off so easily. Shorting out pins right now so it's always on.

I'm using the JJ ECC82 and it sounds much better now. The 6N23 stock sounds still sounds thick.

My question is when I use the 6N23, I can adjust the trim to 13.5V. When I use the ECC82 and set it to 12V, the lowest value I can set to is 16.5V.

What bias voltage should I be using for ECC8?

Thanks for all your help and these excellent tweaks!

[Deleted]

The bias setting 'problem' is because of the much different v/i characteristics of these tubes.
With ECC81 it's probably even worse.
ECC83 (high mu triode) and ECC88/6N23 (also somewaht higher mu) is probably better suited for this amp.

16.5V will work but when clipping levels are reached only one side of the sine wave is clipped causing an effective 'DC component'.
Try adjusting L and R equally to as close as possible to 13.5V (with IRF510 to 15.5V)
You will defenitely hear this distortion and stay away automatically so not a real problem when listening to music and turn the volume lower.
You will probably not be reaching these levels anyway or you'll go deaf !

Mod 4 and 6 have the MOST influence in the sound
try fooling around with mod 4 (between 27 and 120 Ohm)

[lynxo323]

Thank you.

I will stick with 6n23 type tubes then.

Lowest I can go is 16.5V with the ecc82.

I replaced the 30R output resisitor with 68R for now.