Compact Tube Amplifier [130385-I]

By Michiel Ter Burg See also http://www.luidsprekers.org/STA/. The goal was not to build a super Hi-Fi amplifier, but to see if it would be possible to build a reasonable amplifier using standard transformers.

By Michiel Ter Burg

See also http://www.luidsprekers.org/STA/.

The goal was not to build a super Hi-Fi amplifier, but to see if it would be possible to build a reasonable amplifier using standard transformers. This turned out to be feasible.

The sound on different drivers (irrespective of impedance) is as expected from a tube amplifier: detailed by low output volume and smoothly overdriven for large input signals. Although the measured bandwidth was a little disappointing, the overal sound quality is sufficient.

It all fits on a single-sided PCB of 90 by 83 mm with 7 wire bridges, however it is worth the trouble to design a double-sided PCB for the amp.

Discussion (1 comment)

TonGiesberts 11 years ago

Compact valve amplifier (130385-1)

We designed a double-sided PCB with a large ground plane (with a large clearance because of the high anode voltage) to shield the mostly high-impedance components. In addition, a simple current source (using a LM337) is added for the filaments. This way only a simple DC power supply with a voltage higher than -31 V is necessary for the filaments. The size of the heatsink required for the LM337 depends on the amount of extra voltage. If possible keep the temperature drop across the heatsink below 40 degrees (Celsius). The input part of the circuit is laid out as symmetrically as possible. The input transformer is kept at a distance of the output transformer on purpose to avoid magnetic coupling between the two transformers. For the same reason the output transformer is not placed over the other components to avoid modulation in the various parts of the circuit. With JP2 and JP3 the feedback of the output signal to the input can be disconnected (JP2) and also the ground connection can be disconnected to isolate input from output. We tried to keep the PCB surface minimal to keep it a compact amplifier as the original build prototype from the author. That’s why a couple of traces are under the input transformer. So this PCB design should never be connected to a directly to the mains voltage connected power supply. All components are through hole types. The footprint of capacitors C1, C3, C4 and C5 have multiple lead spacing’s to accommodate a variety of types. For the power supply connections and the loudspeaker screw terminal blocks are used. Because of the high anode voltage a terminal block with a lead spacing of 7.5 mm has been used for K3. To maintain the experimental design we placed a pinheader (JP1) to select whether the two secondaries from the output transformer should be connected in parallel or in series. Two jumpers are needed to place them in parallel and only one jumper (in the middle) places them in series.

Power supply (130385-2)

To test the amplifier two supply voltages are needed one for the anodes and one for the heaters. In theory the voltage for the heaters should be two times 13 V (according to the datasheet, or maybe more) plus at least the drop out of the current source. Given a drop out of 4 V of the entire current source (cathode D1 to –Vff) 30 V is the minimum voltage needed. In practice the ripple and mains voltage variations should be taken into account also. For test purposes we used a 2 x 12 V transformer. For the anode more than 10 W is needed for full power. For the heaters also almost 10 W is needed. In the FL series of Block 30 VA is the largest with the same footprint as the output transformer of the amplifier. We chose this one for the heaters .That leaves about 20 W for the anode power supply. An old trick is to simply place another transformer the wrong way round and supply its secondary side with the 2 x 12 V. For this transformer we chose the 18 VA type (also the same footprint). Don’t expect 230 VAC at the output of the second transformer. According of the datasheet the no load voltage for the 18 VA type is 1.22 times higher than the rated voltage. So the true winding ratio is higher 230: (24x1.22). If we turn the transformer around the no load voltage at the output will be lower than 230 V according to this ratio. Instead of 230 V only 188.5 V should be measured. If this supply voltage is maximally loaded, the expected voltage will drop to the ratio of the nominal values in the datasheet. The output voltage will then be 188.5/1.22 is 154.5 V. Since the transformer is not designed to work this way the DC voltage will be considerably less than these calculations. Peak currents for charging the buffer capacitor will produce high internal losses. In our test setup the voltage was a little over 188 VDC (no signal). The no load voltage will be a lot higher and will go beyond 265 VDC (Ö2x188.5). Several capacitors in the amplifier are 250 V type. If this power supply is to be used permanently consider changing the 250 V capacitors to 350 V types. Before the valves warm up, and this takes many seconds, the current drawn from the high voltage is 0 mA! We placed a led across both power suppies. The one across the current source is also an indication that the mains voltage is connected. The one across the high voltage indicates the voltage is present (so be careful) but also acts as an idle load to prevent the voltage getting to high before the valves warm up.

Measurements

Input sensitivity parallel, with feedback 1.7 V (Pmax = 3.1 W)

parallel, no feedback 0.7 V (Pmax = 3.4 W)

series, with feedback 1.3 V (Pmax 1.04 W)

series, no feedback 0.75 V (Pmax = 1.08 W)

Input impedance (1 kHz) with feedback 1.01 kΩ

no feedback 1.065 kΩ

Max. RMS power parallel > 3 W

series » 1 W (series)

Bandwidth parallel, with feedback 28 Hz…6.2 kHz

parallel, no feedback 163 Hz…2.75 kHz

series, with feedback 46 Hz…4.8 kHz

series, no feedback 135 Hz…2.82 kHz

Signal to noise ratio parallel, with feedback > 64 dB

(1 W/8 Ω, B = 22 kHz) parallel, with feedback > 72 dBA

parallel, no feedback > 57 dB

parallel, no feedback > 64 dBA

Harmonic distortion plus noise parallel, with feedback 0.4 %

(1 W/ 8 Ω, 1 kHz, B = 80 kHz) parallel, no feedback 0.9 %

series, with feedback 5 %

series, no feedback 4 %

Damping factor parallel, with feedback 2.76

(1 kHz, 8 Ω) series, with feedback 1.25

Current consumption +Va with feedback 52 mA (no signal, +Va = 188 V)

(parallel) with feedback 59 mA (P = 2.9 W, +Va = 181 V)

with feedback 56 mA (P = 3.1 W, +Va = 181 V)

no feedback 59 mA (P = 3 W, +Va = 181 V)

We also recorded some plots

Plot A (THDvsRFQ_parallel_1W_8ohm_withFB.emf ) shows harmonic distortion and noise as a function of frequency at an output level of 1 W (at 1 kHz) into an 8 Ω load. The two secondary windings are connected in parallel and the feedback jumpers JP2 and JP3 are in place. It’s obvious the transformer really causes a problem delivering power at the higher frequencies.

Plot B (AMPvs FRQ_parallel_1W_8ohm.emf) shows the amplitude as a function of frequency at an output level of 1 W into an 8 Ω load. The two secondary windings are connected in parallel. The curve with the bigger bandwidth is the one with feedback (input level 850 mV). The other trace has no feedback connected. To make the difference more clear the curve without feedback is compensated at 1 kHz to have the same level as the curve with feedback (input level 235 mV).

Plot C (THDvsLVL_parallel_2.emf) shows harmonic distortion and noise as a function of output power into an 8 Ω load. The two secondary windings are connected in parallel. Two curves are shown, one with and one without feedback. Noteworthy is the output power is slightly bigger no feedback is applied. Maximum power is 3.4 W (THD+N = 9.2 %) without feedback and 3.1 W with feedback (THD+N = 6.9 %).

Plot D (THDvsLVL_series_2.emf) shows harmonic distortion and noise as a function of output power into an 8 Ω load. The two secondary windings are connected in series. Two curves are shown, one with and one without feedback. Here the output power is also slightly bigger if no feedback is applied. Without feedback maximum output power is 1.04 W (THD+N = 6.4 %) and 0.99 W (THD+N = 5.1 %) with feedback. As could also be observed in the previous plot output power decreases when the in put level is increased after the maximum power has been reached. Bothe curves fold back.

Plot E (FFT_parallel_1W_8ohm_1kHz.emf) shows a Fourier analysis of a 1 kHz signal (fundamental suppressed) at 1 W in to an 8 Ω load with the secondary windings connected in parallel. Total harmonic distortion plus noise is 0.4 %. Besides the two harmonics that stand out also a large spectrum caused by the ripple of the power supply is present. The two main harmonics at -64,2 dB and -48,1 dB are responsible for a THD of 0.4 % which corresponds with the measured value. Looking at this plot however the amplifier will surely benefit from a DC regulator for the anode supply.

Bill of materials amplifier

Resistor

R1,R3,R6,R8,R9,R11,R12,R15,R16,R19,R25 = 1 kΩ, 0W6, 1 %, 350 V

R2,R14 = 100 kΩ, 0W6, 1 %, 350 V

R4,R17 = 680 kΩ, 0W6, 1%, 350 V

R5,R18 = 10 Ω, 0W6, 1 %, 350 V

R7,R13 = 47 Ω, 0W6, 1 %, 350 V

R10 = 220 Ω, 0W6, 1 %, 350 V

R20,R21 = 47 kΩ, 0W6, 1 %, 350 V

R22 = 2.7 kΩ, 0W6, 1 %, 350 V

R23 = 470 Ω, 0W6, 1 %, 350 V

R24 = 10 kΩ, 0W6, 1 %, 350 V

R26 = 4.7 kΩ, 0W6, 1 %, 350 V

R27,R28 = 8.2 Ω, 0W6, 1 %, 350 V

P1 = 10 kΩ, 0W15, 20 %, trimmer, top adjust

P2 = 1 kΩ, 0W15, 20 %, trimmer, top adjust

Capacitor

C1,C3,C5 = 100 nF, 250 V, 5 %, MKP, lead spacing 5/7.5/10/15/22.5 mm

C2,C8 = 100 nF, 100 V, 10 %, MKT, lead spacing 5/7.5 mm

C4 = 47 nF, 250 V, 10 %, MKP, lead spacing 5/7.5/10 mm

C6 = 10 µF, 100 V, 20 %, lead spacing 2.5 mm, diam. 6.3 mm

C7 = 10 µF, 250 V, 20 %, lead spacing 5 mm, diam. 10 mm

Transformer

TR1 = FL 14/6 (2x115V prim./2x6V sec., 14 VA)

TR2 = LTEI19/KD-0703, Audio transformer 1:5 (Conrad # 515701)

Semiconductor/Valve

V1,V2 = PCL86 (Conrad # 130632)

D1 = 1N4007

IC1 = LM337

Other

K1,K5,JP2,JP3 = 2-way pinheader SIL, 2.54 mm spacing

K2,K4 = PCB terminal block, 2way, pitch 5 mm

K3 = PCB terminal block, 2way, pitch 7.5 mm

JP1 = 4-way pinheader SIL, 2.54 mm spacing

JP1-JP3 = jumper 2.54 mm

V1,V2 = Socket, noval, 9-pin, keramic, PCB (Conrad # 120529)

Heatsink for IC1 30 K/W (Fischer Elektronik SK 12 SA 32)

Misc.

PCB 130385-1 v1.0

Bill of materials power supply (our test setup, no PCB)

Resistor

R1 = 270 kΩ, 0.5 W, 5 %, 350 V

R2 = 10 kΩ, 0.25 W, 5 %, 250 V

Capacitor

C1…C4 = 4.7 nF, 400 V, 10 %

C2…C8 = 47 nF, 100 V, 10 %, ceramic

C9 = 100 µF, 350 V, 10 %

C10,C11 = 1000 µF, 50 V, 10 %

Semiconductor

D1…D4 = 1N4007

D5…D8 = 1N5819

D9,D10 = low current led

Other

F1 = 400 mA/T (115VAC mains) or 200 mA/T (230 VAC mains) plus fuse holder

F2 = 750 mA/T plus fuse holder

K1,K2 = PCB terminal block, 2way, pitch 7.5 mm

K3 = PCB terminal block, 2way, pitch 5 mm

TR1 = FL 30/12 Block (2x115V prim., 2x12 V sec., 30 VA)

TR2 = FL 18/12 Block (2x115 V prim., 2x12 V sec., 18 VA)

130385-1-3.jpg (1352kb)
130385-compacte-buizenversterker-schematic-v110.jpg (1236kb)
130385-2-voeding-compacte-buizenversterker-schematic-v110.jpg (997kb)
Topoverlay of the PCB of the amplifier (130385-1 v1.1) (59kb)
Top copper and bottom copper of the PCB (130385-1 v1.1) (40kb)
Measurents of our prototype (23kb)

kay 10 years ago

First attempt in making tube amplifier. Can some one explain in plain English as how to set the bias....should I connect speakers before measurement and do I measure Milli volts or Milli ampere and across which resistor? Thanks in anticipation of a reply.

oberwolf 10 years ago

Hi, The Diagramm for the Power Supply is wrong. Its says K2 for the Va and K3 for Vff. On the main PCB K2 ist the Connector for the Loudspeakers. So K2 on the Power Supply is real for the k3 on the Main PCB and K3 on the Power Supply is real for the K4 on the Main PCB, Please confirm if this is right. Wolfgang