the matrix amp
TRANSCRIPT
ETF.03
Steve Bench
The Matrix Amp
1
Talk Outline
This file gives an overview of the presentation to be given at the workshop. One of the MatrixAmps (7695) has been constructed for ETF2003. It is typical of the “matrix sound” (whateverthat means). However, it is slightly less polished than some other members of the family. This isdue in part to the use of power transformers (!) for input and output audio transformers, as wellas a desire to highlight the characteristics of the matrix amplifier. That is, it sounds like it is aMUCH more powerful amp than it is, it is inherently inexpensive, and can be very compact andlight. The 7695 amp is small (200cu inch), light (7 pounds) and cheap (about 100 USD), yet pro-duces some reasonably impressive sound. This talk discusses the matrix amp and some relatedtopics of interest to the audio community.
Prepared by Steve Bench [email protected]
2
Overview of the Matrix amp
• Background
• How does it work
• Refer to GEN_MAT1I (page 9). For illustrative example, assume the amplifiers are true voltagesources (i.e., 0 ohms) and the transformers have a 10:1 turns ratio (100:1 impedance ratio). Youcan look at this as a 1k impedance on each half of the primary translating to 10 ohms on eachhalf of the secondary for the tapped transformer and 1k to 10 ohms for the untapped transfor-mer. Now, feed a signal into the left channel only. GEN_MAT2. Since the amps are defined as vol-tage sources, that means that point C has 0 volts on it, half at point B, and the other half A-B.(Autoformer action, it doesn’t matter that effectively there is a parallel transformer from B toC). The vectors on the secondaries then show 0.5V from D to E and another in phase 0.5V fromE to ground. Since the left speaker is connected D to ground, 1 volt is produced. Note that since10V from the source translates to 1 volt at the load, the same 10:1 voltage ratio is produced bythe matrixing system. Note that the vectors subtract for the right speaker (the heads of thearrows face each other) so no sound is produced in the right speaker, which is what is desired.GEN_MAT3 (page 10) describes the system with the right channel driven.
• Limitations (historical)
• Crosstalk limited
• Since one side needs to be “inverted”, was deemed useable only with ceramic/crystal phonostages where the inversion was accomplished at the cartridge.
• One transformer has DC flowing in it. (parafeed was apparently never a consideration.)
• Speaker connections are “odd” in that one side + is ground. This would have confused a lot ofpeople in a “general purpose” amp.
• Effects of source resistance on crosstalk and performance were more difficult to envision, sofeedback was sometimes used as an attempt to “fix” this problem. In this topology, it madethings worse.
• SPICE etc tools not available.
• Modeling (SPICE)
• Basic model is shown in ETFMAT1.TXT. (Hint: if you rename this to ETFMAT1.CIR, it will open upin AIMSPICE directly and allow you to do an AC analysis.)
• SPICE clearly shows the limitations of this topology. See RES1 and RES2 (page 17).
• Overcoming the limitations
• Basically, one limitation to remove is the somewhat limited midband channel separation. Thiscan be done by adjusting the turns ratio of the second transformer, adding a small resistor inseries with the feed-in of the second transformer, and placing a capacitor across the first trans-
3
former to cancel the effects of leakage inductance. (ETFMAT2.TXT, RES3, RES4; both page 18). Apractical consideration that would be nice is to keep the transformers the same. But there is DCin one and not in the other. This is easily fixed by parafeeding the second transformer. We willdiscuss how much leeway you have in selecting these components later.
Characteristics of the Matrix amp
• Apparent power
• One of the characteristics that everyone who has built one of these amps or listened to it hasconsistently noted is that the matrix amp SEEMS to be a much more powerful amplifier than itmeasures. Why? Compare 50C5DIS (page 14) with D_1060_2 (page 16). In particular note thatfor a “traditional” amplifier with no global feedback, the difference between 1% and 3% distorti-on is about 2dB. (50C5DIS). With the matrix, this is extended to about 3dB. This means that dis-tortion is growing more slowly , and this translates to a perception of greater power. Also, com-pare RES3 top curve with RES4 top curve. As both channels are driven the bandwidth extends.Now compare D_1060_1 with D_1060_2. With both channels driven, the available power AT ANYDISTORTION LEVEL increases by about 2dB. These 3 effects combine to cause the matrix toappear more powerful than its rating.
• Note what happens when both channels are overdriven. So long as the signal is mostly com-mon, the channels go into a “distortionless” Class AB operation. (There is no signal in the“second” transformer – more on this later). This actually provides additional power and explainsthe 2dB measured difference in PO.
• SE sound
• Examination of the schematics show that each “channel” is a SE amplifier that is somehowmagically combined in the output transformer primary and separated in the output transformerssecondaries. The amps SOUND like SE amplifiers.
• Components surrounding the amp
• Note RES5.GIF. Here, I have purposely used insanely undersized parafeed components. Yet,compare RES4 with RES6. They’re the same! For low frequencies, where channel separation isunimportant, there is NO signal in the lower transformer. This has an interesting ramification ifyou think about it. We are always trying to minimize the number of components in the signalpath. Here’s a two channel amp that depends on only 1 transformer for its character. Note theschematic 7695MAT (page 11). The amp you’re listening to has a 1.5Hy parafeed choke! I did thatas an illustration of how insensitive the topology is to those components.
4
• Why pentodes?
I’ve built matrix amps with “all 4 permutations” of driver and output. Here’s the subjective“quality” of the resulting sound:
Discussion…
Matrix amp discussion
• Inherent shortcomings
• Does not like speakers with high order crossovers. Compare RES7 with RES8. This shows thedetailed response of one of the channels with both driven. The shown channel is loaded resisti-vely, but the NON-SHOWN channel is loaded with a complex load (LCLC along with R).
• Performance appears to be sensitive to power supply quality. (moreso than conventional SE)
• Inversion in one channel requires either input transformer or an “odd” stage.
• What are additional uses for this topology
• This topology is inherently smaller, and lighter than “conventional” amplifiers. For situationswhere several amplifiers are needed (e.g., home theatre) this would seem ideal.
• Small side installations.
• Why wasn’t the Matrix originally successful?
• Since there were no “EDA”, getting good performance was not necessarily obvious.
• Topology regarded as a “cheap fix” when more-complex-is-better thought prevailed.
• See also limitations listed above.
5
Driver Output Resulting Sound
Triode Triode fair
Pentode Triode much better
Triode Pentode worst
Pentode Pentode best
Overview and review of the constructed Matrix amps
• YL1060
• First constructed matrix. Is unusual in a couple of respects:
• Has “blend / anti-blend” control
• 2 tubes for a stereo amp. Both dual tetrodes.
• matrix
• Also the distortion curves show one channel and both channels driven.
• KT88
• Follow on from above. More conventional. This has been constructed by several people. Likesto be played loud!
• Amp and PS schematics shown.
• 6MB8 input stage (triode-pentode… one of the triodes is used as the “inverter” the other oneis in the power supply as a shunt regulator).
• Some people have reported slight constriction at low levels. See 813 amp.
• 813
• Big brute version; 60 or more watts per channel.
• All power supplies regulated. Supplies somewhat “unusual”.
• Originally constructed with same front end (6MB8) as KT88 version, later changed to 6GX7(slight pin change but NO circuit value changes).
• It is very difficult to overdrive this amp.
• With 6GX7 front end, there is no low level constriction at all.
• Complex, but good. Note that the distortion at 1 watt is about 0.03%.
• 7695
• Very small, very light, inexpensive.
6
Interrelated Topics
• Feedback
• Characteristics of feedback
• When is it “acceptable”
• Local vs global
• Feedback in pentode stages vs feedback in triode stages
• Is an amps sound determined by the driver or the PA stage device?
• Compare D_813MT2 and D_813MT3 (page 15).
• Discussion
• To probe further
• Discussion
7
9
10
A A
B B
C C
D D
E E
44
33
22
11
44
44
55
33
7695
7695
6GU
56G
U5
110V
120V
152V
150V
148V
12V
Yel
Blk
Vio R
ed
Grn R
edB
rn
Blu
Not
e re
vers
e ho
okup
left
/rig
htLe
vel
12V
12.9
V
12.9
V
148V
138V
138V
45V
45V
Yel
Yel
Blk
Blk
Vio
VioR
ed
Red
Grn
Grn
Red
Red
Brn
Brn
Blu
Blu
(c)
Cop
yrig
ht,
2003
Ste
ve B
ench
. A
ll R
ight
s R
eser
ved.
Spk
r Ja
ck
Spkr
Jac
k
138V
128V
0.4V
0.4V
Not
es:
Sen
siti
vity
: 26
5mV
at
J1,2
= 1
wat
t.Z
out
= 4
ohm
s. I
nten
ded
for
8 oh
m s
peak
ers.
Am
veco
/To
lem
a to
roid
s av
aila
ble
from
Dig
i-K
eyL1
is
Ham
mon
d.C
3 is
non
pol
ariz
ed,
not
elec
trol
ytic
Re
spo
nse
: -.
7d
B a
t 20
Hz,
-.6
dB a
t 20
kHz.
-3d
B a
t 40
kHz.
Dis
tort
ion:
Abo
ut 1
% a
t 2.
5WM
ax o
utpu
t =
3.5W
per
cha
nnel
.S
elec
t m
atch
ed 7
695
and
mat
ched
6G
U5
7695
Mat
rix
A
7695
Mat
rix
Am
plifi
er
A
11
Mon
day,
Aug
ust
04,
2003
Titl
e
Siz
eD
ocum
ent
Num
ber
Rev
Dat
e:S
heet
of
Left
+
Left
-
Rig
ht+
Rig
ht-
o oooT5
Am
veco
115
+11
5:9+
9 50
W1
5
26
3 4
7 8
J3 115V
S1
Pow
er F1
1.5A
SB
D1
FR
307
D2
FR
307
D3
FR
307
D4
FR
307
D5
FR
307
D6
FR
307
t
RT
1C
L40
R26
68 5
W+
C11
1000
u 16
0V
+
C10
1000
u 16
0V
+C
6
1000u 160V
+C
7
1000u 160V D7
12V
5W
+C
1210
00u
16V
R20
10 3
WR
2110
3W
R25
100k
2W
R24
100k 2W
+
C5
2200
u 25
V
+
C4
2200
u 25
V
R19
5.6
5WR
2233
0 1/
2WR
231.
6k 1
/2W
+C
8
33u 160V
+C
9
33u 160V
o oooT1
Am
veco
115
+115
:22+
22 3
.2W
15
26
3 4
7 8
o oooT3
Am
veco
115
+115
:22+
22 3
.2W
15
26
3 4
7 8
J1Left
In
J2 Rig
ht I
n
R2
50k
R13
50k
2,
V2
6GU
55
6
73
4
1
2,
V3
6GU
5
5
6
73
4
1
R5
1k
R15
1kR9
75
R10
75
R6
20 R14
20
R3
30k
R16
30k
V1
7695
9
16
74
5
V4
7695
9
16
74
5
C1 1u
200
V
C2
1u 2
00V
R7
330k
R12
330k
R4
1k
R17
1k
R8
10 3
WR
1110
3W
R1
100k
R18
100k
o oooT
2A
mve
co 1
15+
115:
7+7
50W
15
26
3 4
7 8
o ooo T
4
Am
veco
115
+11
5:7+
7 50
W
15
26
3 4
7 8
C3
30u 100V
L1 1.5H 200mA
12
13
14
15
16
17
Top: Res1.gif Bottom: Res2.gif
18
Top: Res3.gif Bottom: Res4.gif
19
Top: Res 5.gif Bottom: Res6.gif
20
Top: Res 7.gif Bottom: Res8.gif
Measurements of the ETF.03-Matrix Amp (by Christian Rintelen)
Both channels driven into clipping (16 ohms) – noteassymetrical clipping
21
Both channels driven (reference tone: square wave equivalent to 1 kHz with 5% THD into 16 ohms)
Ω 20 Hz Ω 50 kHz≈ 100 Hz ≈ 1 kHz
2222
Phase behaviour (sine wave; reference: 1 kHz, 5% THD into 16 ohms)
Ω 20 kHz Ω 30 kHz≈ 40 kHz ≈ 50 kHz
Ω 10 kHz Hz Ω 20 kHz
Christian Rintelen’s measurements of Steve’s ETF.03-Matrix amp:
1. Output:Both channels driven into 16 ohm load; 1 kHz into visible clipping:left channel 5,2 V RMS = 1,7 W, right channel 5,4 V RMS = 1,8 W
2. Output vs. distortion(bear in mind please that the el-cheapo ETF.03-Matrix amp is uses toroid power transformers asinput and output transformers!!) both channels driven into 16 ohm load, all power levels measu-red at 5% THD200 Hz: L = 2,0 W, R = 1,9 W1 kHz: L = 2,0 W, R = 1,9 W10 kHz: L = 1,8 W, R = 1,8 W20 kHz: L = 0,4 W, R 0 0,6 W
3. Distortion vs. frequency(input: sine wave referenced to 1 kHz into 16 watts (both channels driven) and 5% THD). Outputreferenced in dB to 1 kHz input signal = +/- 0dB
left channel right channelTHD output THD output
20 Hz 56% (!!) - 1,35 dB 55% - 1,41 dB50 Hz 5,6% + 0,2 dB 6,6% + 0,2 dB100 Hz 4,53% 0 dB 5,6% 0 dB1 kHz 5% 0 dB 5% 0 dB10 kHz 9,1% + 0,4 dB 8,9% + 0,4 dB20 kHz 22% + 0,6 dB 21% + 0,6 dBS
23
Steve Bench’s measurements of his 813 Matrix Amp
(8 ohm load, both channels driven)
Power = 1 watt20Hz 0.05%100Hz 0.04%1kHz 0.035%5kHz 0.045%10kHz 0.08%20kHz 0.09%
Power = 10 watt20Hz 0.65%100Hz 0.45%1kHz 0.4%5kHz 0.6%10kHz 0.7%20kHz 0.9%
Power = 20 watt20Hz 1.0%100Hz 0.8%1kHz 0.9%5kHz 1.1%10kHz 1.6%20kHz 1.6%
Power = 50 watt20Hz 6% (on verge of saturating the iron)100Hz 2.5%1kHz 3.0%5kHz 3.5%10kHz 4.0%20kHz 4.1%
Power = 60 watt20Hz 10% approximately ... saturation evident100Hz 4%1kHz 4.5%5kHz 4.5%10kHz 5%20kHz 5.2%
24
252525
Rows 1 and 2 show the 60W 813 Matrix amp (measure-ments next page)The picture to the left shows the KT 88 Matrix amp withthe 813 for comparison…
Some pictures of other Matrix Amps
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1(c ) Copy r igh t , 2003 S teve Bench. Al l Rights Reserved.
Left Spkr
Right Spkr
Interconnect
+750VDC
Set idle currentto 100mA ( 1 voltacross 10 ohm.)
+787V
+772V
+772V
-60VDC_aboutBias voltage is
200V
200V
150V
Volume
1.3V
1.3V
Note : Po la r i t y o f R igh t Input connect ion to transformeris inverted with respect to left input.
1V
1V
11V
11V
1V
1V
7V
7V
On Front Panel
C
813 Matr ix Ampli f ier
A
1 1Monday, June 02, 2003
813 Matr ix Amp
Tit le
Size Document Number Rev
Date: Sheet of
Left-8
Left-0
Right-0
Right-MonGround
Right-8
Left-Mon
+750VDC
+10V-1+10V-1Rtn
+10V-2Rtn+10V-2
Bias2
Bias1
+200V
+800VDC
Ground
+150V
813V1cap
34
1 7
5
o
o
o
o
o
o
T2Lundahl 1660
9 21
8
18
13
16
10
72
53
4
o
o
o
o
o
o
T6Lundahl 1660
9 21
8
18
13
16
10
72
53
4
R7 10 1
% w
w
R2100
R10100
T1 Lundahl 1620PP4
25
73
16
81112
13
14
1516
17
18
19
2021
22
23
2425
26
C3
20u 100V NP
813V3
cap
34
1 7
5
C1
.1u 1kV
R8 10 1
% w
w
R4N.U.
R13N.U.
R16 10 3W
R18 10 3W
R15 150k 5W
R1 150k 5W
V26MB8_P
6
7
8 4 5
9
V46MB8_P
6
7
8 4 5
9
R17
100
R3 100
L1Lundahl 1673-10H
1
36
7
T4 Lundahl 1620PP4
25
73
16
81112
13
14
1516
17
18
19
2021
22
23
2425
26
C4 .1u 1kV
C7 10nF 100V
C2.1u 1kV
C6 10nF 100V
C5.1u 1kV
R9100 1%
R11100 1%
T3 tr_Lundahl15401
2
3
4
E
5
6
7
8
T5tr_Lundahl1540
1
2
3
4
E
5
6
7
8
R650k
R1250k
R51k
R141k
J1Left Input
J2 Right Input tR T 2 KC024L (CL101)
tR T 1 KC024L (CL101)
I monitorI Monitor
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1(c ) Copy r igh t 2003 , S teve Bench. Al l Rights Reserved.
+750V+800V
+210V+200V+150V
150VPS LED200VPS LED
ScreenSupply LEDPlateSupply LED
HV LED
HV and LVPS Chassis
Neg, Bias and FilamentChassis
AC Line
AC Line
Power Swi tch
1 +800V2 N.U.3 Ground4 +200V5 Bias26 Mon Right7 +10V-18 +10V-1Rtn
9 +750V
12 +150V
10 N.U.11 Ground
13 Bias114 Mon Left15 +10V-216 +10V-2Rtn
L Spkr +L Spkr -
R-Spkr +R Spkr -
AmpChassis
Input Transformers
and Volume Contro l
on Front Panel
Indicators
on Front Panel
+210VGroundUM81 Fi l+UM81 Fi l -Mon LeftMon Right
-100VGround
1 -100V2 Ground3 Bias14 Bias 25 Ground6 +10V-1
7 +10V-1Rtn8 +10V-2
9 +10V-2Rtn10 -100VSupplyLED
11 UM81 Fi l+12 UM81 Fi l -
I_Monitor LI_Monitor R
1V = 100mA
LEDs on front panel.
813 Matr ix Amp A
Module Interconnect ions etc.
A
1 1Friday, May 16, 2003
Tit le
Size Document Number Rev
Date: Sheet of
Left Input
Right Input
Green
Green
Green
Green
Yel low
Yellow
Yellow
Blue
4.7k
2k 2k
5 .1V 250mW
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1(c ) Copy r i gh t 2003 S teve Bench. Al l Rights Reserved.
+750V
+149V
+149V
+212V
+212V
+600V
+600V
+700V
+700V
+800V+950V
Note: B+ Adj. range:about 750V to 840V
Note: LVPS on same chassisbut shown on i ts own schematic.
RT1-RT4 are Inrush Current L imi tersKC009 (CL90) parts.
Notes:HV PS and LV PS bui l t on same chassis.Nega t i ve (b i as ) and Amp Fi lament supply on separate chassisAmpli f ier bui l t on separate chassisL E D s a n d Level "Magic Eyes" On front panel
T h i s symbol indicates connect ion wi th in same chassisT h i s s y m b o l i n d i cates interconnect to di f ferent chassisT h i s symbol indicates Reference Voltage used in
other p laces on the same schemat ic page
B+Adj.
813 Matr ix Amp B
HV Regulators
A
1 1Friday, May 23, 2003
Tit le
Size Document Number Rev
Date: Sheet of
To Plate Supply LED
+800V Output
To HV LED
-100V Reg Input
+750V Output
To Screen Supply LED
+150V
+150V
+150V
+210V
+210V
+210V
117V
400-
0-40
06.3V
T1 Hammond 278CX
D1 UF1007 D2 UF1007
D8 UF1007D7 UF1007
D3 UF1007 D4 UF1007
D5 UF1007 D6 UF1007
+C2
1500
u 25
0VR6
82k
2W
+C4
1500
u 25
0VR13
82k
2W
+C6
1500
u 25
0VR17
82k
2W
+C7
1500
u 25
0VR18
82k
2W
+C8
1500
u 25
0VR20
82k
2W
D10
5.1V
250
mW
V1 6080
1
23
4
567
8
R2 68 5W
R5 68 5W
R1 1k
R7 1k
R8470k
V2 12AX71
2
3
4 5
6
7
8
9
R3
300k
1%
R9
300k
1%
R1550k 2W ww
R21
249k
1%
+ C147u 450V
R4
180k
3W
R10
180k
3W
D9
5.1V
250
mW
+ C547u 450V
C3
22n
1kV
R12 220 5W
R16 1k
R19 1k
R11 220 5W
R14 470k
V4 12AX71
2
3
4 5
6
7
8
9
+ C1247u 450V
R22
180k
3W
R25
180k
3W
D11
5.1V
250
mW
+ C1447u 450V
117V
125-0-125
6.3
T2 Hammond 261E6
F18A
J2Jack for -Regulator supply
J1Power (Line cord)
S1Power
tR T 2
tR T 1
tR T 3
tR T 4
LP1
120v 75w
V360801
23
4
567
8
C9
.1u 250V
R28 100k
R27 100k
C11
.1u 250V
R23300k 1%
R24300k 1% C13
22n 1kV
R26249k 1%
C10.1u 250V
250VAC to .. . .
. . .LVPS Rect
+210V Ref (LVPS)
+150V Ref (LVPS)
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1(c ) Copy r igh t 2003 , S teve Bench. Al l Rights Reserved.
Input t ransformers and loudness on front panel
Matr ix Amp A
Alternative Input - as built.
A
1 1Friday, May 16, 2003
Tit le
Size Document Number Rev
Date: Sheet of
To 6MB8 gr id resistor (R5)
To 6MB8 gr id resistor (R14)
50k 13ktap
tr_Lundahl15401
2
3
4
E
5
6
7
8
20k
20k
20k
2k470pF SM
Left Input
2k
1u 200V
2k
tr_Lundahl15401
2
3
4
E
5
6
7
8
470pF SM 50k 13ktap
2k20k
Right Input
20k
1u 200V
20k
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1(c ) Copy r igh t 2003 , S teve Bench. Al l Rights Reserved.
250V
Note tha t the Zener str ing only conducts duringlow or no loads (startup) and with normal load,i t is not conducting.
Notes:HV PS and LV PS bui l t on same chassis.Nega t i ve (b i as ) and Amp Fi lament supply on separate chassisAmpli f ier bui l t on separate chassisL E D s a n d Level "Magic Eyes" On front panel
T h i s symbol indicates connect ion wi th in same chassisT h i s s y m b o l i n d i cates interconnect to di f ferent chassisT h i s symbol indicates Reference Voltage used in
other p laces on the same schemat ic page
250V
250V
+210V
+200V
+150V
-97V
-97V
-100V
+211V
+201V
+151V
+0.6V
+0.6V
+0.6V
Note: Al l three 2n2583 are TO-66 package and shouldbe hea t sinked using a T0-66 Heat sink of 10 degreeper watt or better heat sink.
T h e 210V supply dr ives UM81 indicators and isinternal reference used in HVPS.
T h e 2 0 0 V s u p p l y provides plate supply for 6MB8.
T h e 1 5 0 V s u p p ly provides screen grid voltage for6MB8, and internal reference for HVPS.
813 Matr ix Amp A
LV Regulators
A
1 1Saturday, May 24, 2003
Tit le
Size Document Number Rev
Date: Sheet of
-100V (also used on HVPS)
Gnd (a lso on HVPS)
+210V
+200V
+150V
150VPS LED
200VPS LED
-100V Ref
-100V Ref
+210V
+210V
-100V Ref
-100V Ref
D1 UF1007
D5 UF1007
D2 UF1007
D3 UF1007
R1 620 10W
+C233u 350V
L1 1Hy 75mA
+C347u 350V
D4 28V 5W
D8 28V 5W
D10 28V 5W
D16 28V 5W
D13 28V 5W
D7 28V 5W
D9 28V 5W
D11 28V 5W
D14 28V 5W
D17 28V 5W
R15300k 1 /2W
C70.1u 250V
Q 1 2N3583
Q 2 2N5656
R647k 1 /2W
R3 13k
+C4
22u 250V
R4 13kR2 10 +C1
22u 250V
R5100k 1%
R747.5k 1%
D6 UF1007
Q 3 2N3583
Q 4 2N5656
R1247k 1 /2W
R9 20k
+C6
22u 250V
R10 20kR8 10 +C5
22u 250V
R11100k 1%
R1347.5k 1%
D12 UF1007
D18 UF1007
R14 10
R2015k
R1915k
Q 6 2N5656
R1875k 1%
+C822u 250VR16 15k
+ C9
22u 250V
R2147.5k 1%
Q 5 2N3583
R17100k 1%
D15
5 .1V 250mW
D19
5 .1V 250mW
250VAC f rom HVPS
250VAC from HVPS
+210V
+150V
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1
Sens
itivit
ySe
nsitiv
ity
S e t sens i t i vity for "ful l scale" = 20WNote : Max sens i t i v i t y i s ful l scale = 8 watts.
( c ) Copy r igh t 2003 , S teve Bench. Al l Rights Reserved.
Matr ix Amp A
Signal Indicators
A
1 1Friday, May 16, 2003
Tit le
Size Document Number Rev
Date: Sheet of
Left Mon
Right Mon
+210V
UM81 Fi l+UM81 Fi l -
Ground
V1UM81
9
1
2 4 5
7R31MegC2
0.1u 50V
R21.8k
C11u 200v
D21N5818
D11N5818
R1 510k
V2UM81
9
1
2 4 5
7R61MegC4
0.1u 50V
R51.8k
C31u 200v
D41N5818
D31N5818
R4 510k
A
A
B
B
C
C
D
D
E
E
4 4
3 3
2 2
1 1
AstecLPS6312V 6ASMPS
L
N
E
123
456
+
-
Astec3
4
N
E56
+
12V 6A
-
LPS63
12
SMPS
L
(c ) Copy r igh t 2003 , S teve Bench. Al l Rights Reserved.
+
N e g a t ive and Fi lament Supply Connector
-170V
About 16.5V
SD
G
leads short,Keep these
-85V -85V
-84.4V
-100V
-100V-104V
Optical ly Coupled
Note: IRF9630 is a TO-220 partmounted on a heat sink of10 degree per watt or better.
Bias1Adj.
Bias2Adj.
Bias vol tage adjustment rangeis about -35 to -80V.
A s tec suppl ies are adjusted for about -11.5Vand this al lows about 1.5V drop in the wir inga nd connectors to the 813 sockets.
Note that al l the f i lament suppl ies areunreferenced at this end.
-3.4V
-6.8V
1 2 V 100mA fan moter to cool Astec suppl iesRuns on about 9V to keep noise down.
N o te Schottky Rectif iers
813 Matr ix Amp A
Negative Reg and Fi l Supply
A
1 1Sunday, May 25, 2003
Tit le
Size Document Number Rev
Date: Sheet of
12 UM81 Fi l -
10 -100VSupply LED9 +10V-2Rtn8 +10V-27 +10V-1Rtn6 +10V-15 Ground4 Bias2
1 -100V Reg output2 Ground3 Bias1
11 UM81 Fi l+
117V
125-0-125
6.3
T1 Hammond 261E6
F11.5A SB
J1Power
D4 3A 30V
D5 3A 30V
+C5
15000u 15V
+C6
15000u 15V
M1 12VDC Fan
R16 22
C7.47u 25V
D1UF1007
D3UF1007
R3 100 1W
+ C233u 250V
+ C3
1800u 200V
R5220k 1 /2W
Q 1 IRF9630R4 10
D2 12V 250mW
R72k
R615k
R815k
+
C4 22u 200V
R1 68k R2 12k
Q 2 MPSU60Q 3
MPSU60
R1122k
R97.5k 1%V1 5783
2 31
R1040.2k 1%
D6
Whi te LEDD75 .1V 250mW
+ C110u 160V
R1215k
R17
50k
WW
R2136k
C8
.47u
100
V
Q 4
MPS
U10
R18100
R1336k
R1415k
R19
50k
WW
R2236k
C9
.47u
100
V
Q 5
MPS
U10
R20100
R1536k