TRAM w.

A very long journey into making a line stage amplifier

Next to my vdh Colibri cartridge and vdh phono stage, this 6N6P w.o.t. line stage is the best thing that has entered my hifi system ever. Trying out the TRAM with my Audio Mirror monoblocks and later GlowMaster KT88 amplifiers has been a true revelation. I'm afraid to say it even beats my 300B integrated by a small margin in overall dynamics and an almost palpable midrange that reaches out into my listening room.

I had no idea what I was in for when I borrowed the TRAM, but after a few minutes of listening, there was no doubt in my mind that I had to build a similar line stage. Cloning the TRAM isn't particularly difficult having the amp and assembly instruction at hand. The TRAM, as-is, has serious hum problems and I was sure I could handle this by a better layout and a DC filament supply for the 5687 tubes, but things turned out to be not that simple. Read on.

All transformers from SAC Thailand: http://www.sacthailand.com/

5687 tubes

Initially I bought JAN Philips 5687 tubes for my line stage. Michael/DK gave me a pair of National 5687 and
thanks to John/Utah/US, I now have a wide selection of 5687 tubes including a 5687 substitute, the Dutch Amperex 7119.

Left: National 5687. Middle: Amperex 7119 (Herleen, Holland Amperex "PQ"). Right: RCA 5687

Left: Sylvania 5687. Middle: Rytheon 5687. Right: JAN Philips 5687

How it all started
- the TRAM line stage from DIY-HiFiSupply, Hong Kong

Tubes: PSU: 5Z3, 6AS7, ECC83. Amplifying stage: 2 x 5687 JAN Philips.
Interstage trafos: Origin not known. 10k primary, 0-300-600 ohms secondary. Phones from 300 ohms tap.

The TRAM has some lay-out drawbacks. One is the ECC83 placed far away from the 6AS7 tube, though closely interlinked.
Next all tubes are heated by AC and I think this is one reason for serious hum problems. The use of un-shielded signal
cables may pose another problem and based on the experience presented here, I would re-build the entire TRAM. As-is it
simply can't be used with 95 dB speaker systems. 90 dB and a dead-quiet power amp may go, but that's about it.

(Changing filament supply to DC helped somewhat, but the main problem is likely to be the mains transformer being too close to everything).

My 5687 W.O.T. line stage project:

New layout for wot line stage.
Cabinet dimensions will be 250 (D) x 400 (W) x 75 (H) mm.

Schematics

Complete schematics. Kiwame resistors and Obbligato film caps used in most places.
In the final edition, the filament supply for the 5687 tubes will be changed to a regulated 12.6 V supply
with the two 5687 filaments in series.
For low efficiency speakers the above may work, but not for 95+ dB speakers. Still some residual hum.

Please note: The red 100 ohms resistor was not present in the TRAM circuit.
According to Peter Pan - a well know valve guru - this resistor MUST be there to make the 3.3 uF Obbligato work properly.

Valve pin configuration.

Components

Left: Components piling up.
Right: Interstage trafo, 5k prim./0-300-600 ohms sec. Supermalloy core.

Left: Interstage trafo. Right: Mains trafo.

Left: 6AS7 GEC, JAN Philips 5687, ECC83 Electro Harmonix, BRIMAR U52.
Right: 100k, 23 step latter type attenuator volume control, Dale resistors (50 US $).

Out of three attenuators bought from Hong Kong, three has been faulty. The nice looking one above too. Now, a rotary switch used for an attenuator has to be shorting and turning this one up for the first time made the poor test drivers' cone pop out of the magnet gap and the voice coil stayed on the pole-piece... Hmm.... Sorry Hong Kong dealers selling cheap China stuff, I've had it. This was the last time.
I'll be trying a 41 step Acoustic Dimension attenuator - and my never failing 100k ALPS is doing the job until the AD arrives.

Left: Obbligato 100 uF/400 V; this cap is BIG!
Right: Input selector 2 x 4

Cabinet construction

Cabinet made from 2 mm aluminium. Here with red primer spray paint.
In the town where I live, Aarhus, we have a public workshop, Huset (The House), with a metal workshop, where I can bend the aluminium sheets.

Assembly

Power supply caps take a considerable amount of space.

Left: HT section, 6AS7 on top, ECC83 at bottom.
Right: Amplifying stage, 5687 sockets and two 3.3 uF Obbligato caps.
Only half 5687 is used (and heated) per channel.

Time to start thinking about front panel design.

Hum has been a serious problem with this line stage. Having 95 dB speakers I need very low level noise and the original TRAM was really noisy, something that can be read on the web too if you google "tram wot". AC is used for filament supply in the original TRAM and from all similar wot line stage constructions I have found on the web, all have DC filament supply for the (5687) tubes. Having ordered the mains trafo with 6.3 volt for this, I had to use a simple approach rather than 7806/LM317 regulators, thus 2 x 10,000 uF caps as seen in the schematics. This reduced hum considerably and one channel was silent. The other channel still had unacceptable hum levels (I use a 12" Altec 414C right in my face on the workbench while working on the hum problem). Moving wires around at any stage during construction didn't make any reduction in hum at all. Trying true start earthing of all ground connections didn't help either. H(u)mm....

Some good advice on hum problems can be found here at: http://www.vt52.com/diy/tips/tips_hum.htm
(my comments in red)

Causes
1a - the tube is pulling more current than the cathode emission can handle.
Not the problem here. Have tried increasing cathode resistor. No change in hum.
1b - the tube is resonating along with one of the transformers by means of the chassis or wiring.
Not the problem here.
1c - the ripple on the filament or B+ power supply is too high.
Right channel is silent, so filament supply and HT supply should be OK.
1d - the last capacitor in the filament supply is too far away from the tube itself.
Have tried adding caps right on tube pins - didn't help.

Prevention
2a - use a separate transformer for the filament supplies to prevent capacitive coupling between windings.
Have tried external, heavily regulated 6 V supply: No change in hum.
2b - make sure the power supply for the filament has a ripple of no more than 5-10mV
- have to do something here. I measure 55 mV. I'll try 12.6 volts, LM317T regulation, and connect the 5687s in series.
2c - make sure the B+ supply does not have a ripple greater than 20mV
- I measure less than 1 mV, can this really be true?
2d - use star point grounding for all components and inputs/outputs
Done. Still hum.

Damage Control
3a - place a cathode bypass cap of about 1000uF...
Tried, doesn't help.
3b - use battery heating on the filament...
Not tried.
3c - add more capacity to the B+ supply...
Tried. Doesn't help. (+150 uF extra).
3d - remove the ground point from earth/chassis and connect the two by means of a 100 ohm resistor...
Done. Doesn't help.

I have built quite a range to tube line stages and even RIAA amps in the past and I never had serious hum problems - but..... I never had the mains transformer on board, so to speak.
Adding a large sheet of aluminium between the mains trafo and left channel output trafo reduce hum, so maybe I made a bad cabinet layout after all.

Removing the mains transformer calls for some serious redesign measures and 40 cm width may not be enough for placing mains and output transformers with enough distance. I could place the output transformers sticking out on the left side, but I don't want to make this line stage look like a freak, so first of all I have to remove mains transformer and connect via a 1 meter cable. Takes a lot of wires....

Removing the mains transformer helped a lot and it didn't need being meters away. Actually it could stay on top of the chassis, but only turned 90 deg and close to the front. After this the left channel had the same low hum level as the right channel, but still a bit too much. Removing the choke also from the inside to the top behind the mains transformer helped somewhat further and if I ran my Audio Mirror amps at 1 volt sensitivity (currently at 500 mV), I'd probably be where hum level is acceptable. That is, I'm not sure the AM amps are really dead quiet - or the RIAA for that matter. But, sitting 2-3 meters away from the DTQWTs, I don't want to hear any noise at all. Nothing.

So, time to reconsider the whole project. Right now my w.o.t. line stage looks like something out of a junk yard, and while I'm recovering from the many late night hours fighting the hum war - and cleaning the workshop that looks like a battlefield - I'm now at least enjoying Anne-Sophie Mutter playing Bach - without hum in the quiet passages.

2nd cabinet layout

Alu sheets have been ordered for a new version with separate power supply unit. I'm not a fan of having the lower shelves filled with power supplies, but the 5687 line stage is so good, I'll tolerate almost anything.

Before the new cabinets are finished, more experiments are being performed. Looking through the wot line stage constructions that can be found on the web, none of these had the capacitor connecting B+ and cathode resistor. I'm not sure about the rationale behind this capacitor, but removing it is an obvious option.
Before even doing so, I had some kind advice from Peter/Acoustic Dimension on this: In my first preamplifier with the 5687 + OPT, I also used a separate power supply. Still you have to be also careful that the OPT’s don’t pick-up anything from power transformers from other equipment. There are 2 things I would test to see if the sound can improve still further. Remove the 3,3uf Obbligato across the tube and OPT. Use a higher value resistor, for example 1K2 or no resistor at all (but then you have to test the frequency response and see if it is still OK) I will look now and then to your progress!
So, the 3.3 uF cap was removed and frequency response was measured 1) with 3.3 uF in place, 2) without 3.3 uF + 620R load and 3) without 3.3 uF + 1.1k load. Graphs seen below. With the 3.3 uF cab in place we have a rather early roll-off below 100 Hz, which didn't in any way render the line stage low in bass response, but without the cap it surely looks better; within 1 dB from 20 Hz to 20 kHz.

Left: frequency response from 3.3 uF in place and 620 ohms load.
Right: Red = 3.3 uF removed, still 620 ohms load. Blue = 3.3 uF removed and 1.1k load.

So, with the capacitor removed and a 1.1k load on the output transformers ( + a new 15 V toroidal transformer for the regulated 5687 filament supply) it was time for a new audition with my Audio Mirror mono blocks driving the DTQWT speakers. Hum was almost absent from these last tweaks.
Bass performance didn't appear increased in level from this, but the midrange certainly got rid of some minor smearing of detail. Vocals and piano stood out cleaner, more neutral than before and well, another step in the right direction. Next I have to try no load on the output transformers, that is only the 56k from the Audio Mirror amps.

Current schematics:
3.3 uF Obbligato cap removed.
OT load is increased to 1.1k.
12.6 volts regulated filament supply for 5687 tubes (filaments connected in series)

Building two versions of the 5687 line stage

The outcome of all above is two versions of the 5687 line stage. #1 is for my workshop system, and this will be an integrated version like what has been running in my main system for a couple of months now. Two new mains trafos have been ordered with 14 volts outputs for regulated filament supply for the 5687 tubes. #2 will be a version with separated power supply for my living room system. This will have a potted 50 mH choke to make an overall nice appearance.

So, the line stage was completely dismantled - takes ½ hour to ruin many hours of work! A new top panel was made and the whole thing was sanded and the spray work started all over.

New top panel in place and chassis repainted.

Components shaping up once more.

de Luxe WOT

Teflon sockets. These sockets are second to none compared to any other sockets I've ever tried.

Alu sheets shaping up.

All chassis parts after primer paint.

Paints used:
1. Primer spray; quite pricy this one, but good.
2. Spray used for stoves; makes a nice metallic surface, but rather fragile, thus...
3. Matt acrylic coating.

All of the used spray paints dries very fast, which is important as I do it outside and have to keep fingers crossed that no bugs or any wind born garden debris passes by until it's safe to take inside.
I never do high-gloss painting as I do not have the facilities to do so. Takes a serious dust-free environment.

ELMA attenuator

Had an old - but not very much used - 24 step ELMA attenuator set up for 10k,
so all resistors were removed and a new line for 100k was bought. A bit messy
after all the soldering, but it works. Plan is to use this attenuator in TRAM #1.

Every time I have to set up an ELMA attenuator I have to start thinking how it works, so here's for my own recollection:
Resistors:
40R2, 23R2, 40R2, 40R2, 60R4, 82R5, 115R, 165R, 232R, 330R, 464R, 649R, 910R, 1k3, 1k82, 2k61, 3k65, 5k23, 7k32, 10k5, 14k7, 20k5, 30k0.

The attenuator starts with a maximum attenuation of -68 dB and then has 4 steps to -54 dB.
Each additional step is another -3 dB. All in all better low level control.

Attenuator seen from rear.

Building the 5687 #2, June 09

Paintwork finished and mounting components can begin.

Input selector in place and PSU taking shape.

Almost ready for first time turn-on!

Finally, the 5687 line stage, version #2 is playing and measuring the noise level on the CLIO looks satisfying. This doesn't mean the system is not completely free of hum as one of the Audio Mirror amps needs an overhaul, but I'm pleased the 5687 is off the bench for some time. What's left is making front panels, led indicators and a relay switch that allows the preamp to be turned on from the main unit.
Generally the 5687 WOT is sensitive to nearby transformers of any kind, so if you need your preamp to be cramped in with other gear, this may not be the way to go. But the sound.... the best I've ever had from any linestage.

July 09

The 5687 line stage has now been playing for a month and a half - I needed a brake - and a few holidays have been spent finishing the construction, that is: 12 volt relay that allows me to turn on the amp from the main chassis (small trafo and relay i upper right corner of power supply unit). Next the front panels have been veneered and are ready for drilling, routing, lacquer, etc.

Finished with phones output added!

Read about other wot line stage projects found on the web.

Return to intro page.

Aftermath
August 2009

Building amps and speakers is a never ending story.
I had these comments from Peter Pan:

If you use a capacitor between +B and tube cathode you must include a small value resistor in the +B line. This capacitor is basically a bypass capacitor. If you just parallel up capacitors you end up with a new multi-resonant network that includes both capacitors resonances and two new ones. Adding this resistor serves to “direct” the current at higher frequencies only into the new capacitor. The value of this capacitor should be no larger than (or rather should ideally be) the value of the cathode capacitor divided by Mu – 1. The series resistor should be no larger than the anode impedance of valve used at the operating point used divided by ten. The RC timeconstant between resistor and capacitor determines which frequencies are carried by the +B to cathode capacitor and which will flow through the powersupply and cathode capacitor (it is basically a highpass). In order to get the desired results the +B to cathode capacitor needs to be of significantly higher quality than all the others. I mean, when using the caps without the additional resistor. The TRAM omitted this resistor in the schematic and the build units, as the people making the kit did not understand it’s function and hence decided to “improve” the design. The capacitor value is also not quite right. With 100uF cathode bypass capacitor and a 5687 operating into a nominal 10k load it should be around 6.6uF, not 3,3uF.
With the 7119 (E182CC) the gain is higher and the value is slightly lower. I would use with Ck = 100uF: R = 100R, C = 4.7…5.6uF. This places the turnover at around 300Hz, which is where I usually place it.

The philosophy behind this approach can be read from Lynn Olson's lecture on current loops: http://www.nutshellhifi.com/library/ETF.html.
So, 4.7 uF silver/gold cap was inserted between B+ and 5687 cathode. I'm currently running the Amperex 7119 tube, thus 100 ohms and 4.7 uF.

Does it improve the sound? Indeed it does. Not a giant leap, but a significant improvement. My system wasn't exactly lacking detail and transparency before, but in terms of transient quality and harmonic structure I get a subjectively more true presentation. In short: It makes listening easier. Thanks, "PP".
Amended schematics below.

The RC-circuit inserted.

Next...

2 x 3V lithium batteries inserted in signal path and cathode resistor and capacitor removed. RC circuit left as-is.

When I had this suggestion from "Peter", I thought he'd made a typo or something. A battery in series with the signal? Gotta be kidding!

As it happens, our/my “ultimate” version of this concept also uses grid battery biasing (just measure the voltage across the cathode R – I have no notes left – and put suitable batteries in series with the grid, cathode straight to ground. Bypass the battery stack (Lithium Type Coin Cells is what I use) with a silver mica cap of at least 10nF and a 4.7 … 22MOhm resistor.(This is) - not very common because people believe if it is in series with the grid; it is “in the signal path” and if it is in series with the cathode it is not “in the signal path”. So they stick a big, bad, nonlinear rechargeable Nicad Battery in the cathode, where it is exposed to a lot of varying current so this non-linearity can be injected into the input voltage loop (grid to “ground”) and amplified and they are pleased as punch with themselves for having eliminated another evil capacitor ;-).
Instead they could have placed a nice linear primary cell (non-rechargable – lithium is best IMNSHO) in series with the grid, where there is practically no current flow at all (DC and AC will be in nanoampere region, if any) and eliminating any battery discharge noise needs a very small value capacitor...
Keep the 1K Gridstopper as close to the actual gridpin of the 7119. Obviously, negative pole from the battery to grid. Consider using lithium button cells, they have very little added stray capacitance (small size) and will last many years in this application, the ones for LCD wrist watches. They do come in 3V so only two in series….Regards, Peter Pan.

6 V lithium battery stack, 10M resistor and 10nF silver/mica in place, resistor hidden between components.
Right: These small battery sockets are brilliant and can be found for both one and two batteries. Makes implementation very easy.

Frequency sweep from battery bias set-up. Blue = phase. As can be seen, the line stage
inverts phase, thus output transformer inputs can be swapped or speakers connection reversed.

From cartridge to speaker, the signal in my system passes 7 valves, four coupling caps, three transformers, three interconnects, numerous wires, etc., etc. Tweakings have to be radical to make a significant change to the overall sound of the system, but a system is no better than the weakest link. One thing I can say: These lithium batteries are staying!

-o-o-0-o-o-

October 2010
6N6P valves for my WOT line stage

I was looking for some 7119 tubes on eBay when some Russian 6N6P tubes caught my attention (MYCOMPONENT/Ukraine). I bought eight. For the last year or so, I've been using Amperex 7119 tubes outperforming the 5687 valves - to my ears. The 6N6P valve is a Russian equivalent to E182CC/5687 except for different pin layout. 6N6P data here.
6N6P takes minor rewiring, but 12V DC filament supply can stay, connecting both valves' filament in series as was done for the 7119 tubes. Layout for 7119/5687 heats only one half the tube, allowing tube swapping and doubling tube life, which really doesn't matter any more as 6N6P cost is close to nothing. For 6N6P both triodes are heated, although only one used for amplification. 6N6P has a screen between the two triodes, but only using one triode, I didn't make a ground connection.

My 5687/7119 wiring on top. As can be seen, I only heat one half of the tube, allowing tube swapping and doubling tube life.

Right out of the box, these 6N6P tubes serves you another meal compared to 7119. With no changes to the overall schematics, they appear having slightly less gain, but no problem with this.
Where 5687 in this set-up may seem a little hard and bright, 7119 has an expansive, dynamic and maybe an overly lush sound. Amperex 7119 is truly seductive and maybe not the most "honest" tube we can find. 6N6P appears very neutral compared to 7119 and soundstage depth appears excellent, yet at the same time 6N6P seems to pull everything a little towards the center. Sibilance is significantly reduced compared to 7119 and high piano notes appear to die faster. Does the 7119 in reality have a (serious) problem with microphony? 6N6P tube build appears rock solid compared to 7119 and I never tried ring dampers on 7119 to hear what would happen.
After some 20 hours of playing my initial impressions remain and the tendency to pull the sound stage a little towards the center seems gone. This is probably the most neutral sounding valve I have tried yet and it made me go trough my music collection again, picking out flutes, piano, violins, etc., instruments with a high content of overtones and time and again this tube delivers a more neutral sound without the artificial bloom (smear) from the other tubes tried. And it does well on female vocals too...

Re-wiring my workshop WOT line stage for 6N6P tubes left the same impression as above. These tubes appear very neutral and leave even some harsh sounding CDs on the spinner.
One last important thing I've noticed: When I make a change and treble level seems to has gone down - and it really hasn't - then things are moving in the right direction. Main focus is on the midrange where things happen - and treble is just something that is there - not something that attracts attention on its own.

Re-wiring my friend's TRAM line stage from 5687 to 6N6P left the same impression. Simply more hifi in the true sense of the word. My friend's TRAM is driving a 40 wpc 6C33 PP power amp driving a pair of DTQWT speakers.

No need to say more but this: "Louis, I think this is the beginning of a beautiful friendship".

29-03-2012. I had this mail from John Broskie, Tube Cad Magazine:

Hi,
While searching for 5687 prices, I stumbled on to your webpage that detailed your efforts with the WOT 7119/5687/6N6P line stage. Interesting design. Please let me make a few suggestions. The PS reg is not optimal for several reasons: the 6AS7 is overkill, yet suffers from low transconductance; the 12AX7 is both current and voltage starved. Here is how I would do it:

The 6BX7 holds the same pinout as the 6AS7 and draws only 1.5A of heater current (use an Amperex GZ34 or an RCA 5R4—the small one, not the fat one—rectifier to match the 6BX7 envelope). The 10V zener gives the 12AX7 more plate voltage to play with and the 10k & 10µF prefilter RC section greatly improves the performance. The 1N4007 is a protection device for the 6BX7 and it falls out of the circuit under normal operation. Pots often lose contact, so the above setup is safer and gives more subtle control over the output voltage. In SPICE simulations, this regulator circuit gives a 12dB greater PSRR (power supply rejection ratio) than the original circuit. The PS PSRR across frequency graph look good:

By the way, I could not figure out if the hum you encountered was 50Hz or 100Hz. If the former, then the most likely source is a ground loop or AC induction; if the latter, then the PS is to blame. Another by the way, there is an easy alternative circuit for the line stage amp that would yield a tenfold (-20dB) decrease in distortion and a huge increase in PSRR, but I have to get back to work. Regards John Broskie. PS: I plan on reading the rest of your website soon.

Here the combined schematics.

Thanks to John for taking the time to respond to my WOT line stage!
I did have some trouble with the PSU having to swap ECC83 tubes to find one that would draw enough current to make a stable output voltage. However, no big problem and the line stage has worked fine for several years now. Nevertheless I have bought components for implementing John's suggestions and will report asap.
07-05-2012: Design implemented and it just works excellent. Voltage levels rock stable and fully adjustable via 20k potentiometer. Highly recommended should you decide to build the line stage.

I've had a few line stages in for comparison over the years but none that made me even consider changing this WOT line stage. Even the TRAM2 features capacitor coupling and no matter how good caps we use, caps are crap and colour sound. They are a necessary evil and whenever possible we should reduce the number of capacitors directly in the signal path. This doesn't imply that all the other caps in the design are not a part of the performance but most likely to a lesser extent. I'm not sure how important the 10 uF/400 V cap is, probably not a whole lot. For the 1 uF/400 volt I had two spare silver/gold super caps (Jantzen Audio), actually 1 uF/800V, and the 22 uF/800V is Jantzen Superior-Z cap. The 4.7 uF/800V is a silver/gold Jantzen Audio too. The battery bias may be part of the good sound too eliminating the electrolytic cap across cathode resistor.

2013: Having the WOT line stage running my Glowmaster KT88 and Hypex amps for bi-amping I needed to add four XLR sockets to the rear panel, allowing full balanced operation.

When this line stage is finished, it's time to start building a new chassis! It never ends.

Finally, gain? 14 dB.